greetings --Simon Michael, Mon, 07 Jan 2008 07:25:33 -0800 reply
Happy new year all. I've just returned from a visit to Ireland, where I had a wonderful time (visited family, drank guinness in Matt Molloy's, climbed a mountain, stayed in a castle, jumped in a lake to start the new year..)
I haven't been active with Zwiki development for some months, having been focussed on consulting projects, but I am currently working on some sysadmin tasks, which you should know in case there are glitches. I'm moving domains to dyndns and sites to slicehost. Aktiom was good but slicehost looks better; I can run ubuntu gutsy, and I can get eg 512mb ram for $30/mo ($45 with backups). I've just finished configuring apache and breaking up my ancient crufty apache2.conf (hurrah!)
On the new server I hope to be able to host, develop and test more effectively and complete (or burn! :/ ) the all-unicode branch, among other things. We'll see what the year brings for Zwiki and indeed Zope 2, 3, Grok, Plone and all.
I wish you a peaceful and fulfilling 2008! See you online.
greetings --betabug, Tue, 08 Jan 2008 23:54:55 -0800 reply
Happy new year and welcome back Simon! Looks like you had a really good time in Ireland! The sysadmin stuff looks promising. Best wishes for a good 2008 from me too!
darcs patch: 1397 - fix revert by re-adding setVotes. --Simon Michael, Fri, 18 Jan 2008 08:38:14 -0800 reply
On with the patches!
Thanks for this, applied. I had trouble applying it due to a darcs bug, but pulling it from your repo worked fine. It will go live on zwiki.org when I restart that in a little bit.
outage --simon, Mon, 04 Feb 2008 15:11:42 -0800 reply
The server was down this morning due to a failure of memory-limiting and mail notification scripts. Sorry about that.
... --User123, Tue, 05 Feb 2008 04:54:33 -0800 reply
Requirement is like a user from one machine should access the documents uploaeed on zwiki which located on some another machine..(might be on server machine ).Could you give me early pointers.
translation process --simon, Wed, 06 Feb 2008 15:59:19 -0800 reply
FYI: I just cleaned up and proposed a new translation process at http://zwiki.org/I18n . If you're interested in using Zwiki in other languages and maybe contributing a translation or two (it's easy!), I recommend you subscribe to that page to listen in.
darcs patch: replace: subscriberCount, wikiSubscriberCount with: su... --Simon Michael, Thu, 07 Feb 2008 11:40:22 -0800 reply
experimenting --simon, Thu, 07 Feb 2008 11:58:15 -0800 reply
Experimenting to find a sane process here..
darcs patch: replace: subscriberCount, wikiSubscriberCount with: su... --Simon Michael, Thu, 07 Feb 2008 12:00:09 -0800 reply
... --simon, Thu, 07 Feb 2008 12:24:12 -0800 reply
Subject: darcs patch: replace: subscriberCount, wikiSubscriberCount with: su... Message-ID: <20080207122412-0800@zwiki.org> In-Reply-To: <20080207200007.6FF8D352A78@pool-71-107-253-58.lsanca.dsl-w.verizon.net>
Applied.
darcs patch: strip some bottom-quoting from mailins --Simon Michael, Fri, 08 Feb 2008 20:00:25 -0800 reply
not in Google --geon, Tue, 12 Feb 2008 22:55:35 -0800 reply
Before upgrading to 0.60 our zwiki site was well placed in Google. After upgrading we move somewhere to 100.place :-(. Our unicode signs are wrong reprezented on search result on Google. F.e. insted of "í" there is "Ã". I think it might be due to <html xmlns="http://www.w3.org/1999/xhtml" xml:lang="en" lang="en">. Could it be so? Where to change it? Thanks a lot.
applied patch: strip some bottom-quoting from mailins --zwiki-repo, Wed, 13 Feb 2008 02:31:55 -0800 reply
test 1 --simon, Wed, 13 Feb 2008 02:42:27 -0800 reply
test 1
applied patch: more CHANGES updates --zwiki-repo, Wed, 13 Feb 2008 03:07:31 -0800 reply
applied patch: add the nav links to the help page, testing --zwiki-repo, Wed, 13 Feb 2008 03:23:07 -0800 reply
not in Google --simon, Wed, 13 Feb 2008 04:16:58 -0800 reply
Hi geon. If that's so, you can change it in one or more templates in ZWiki/skins/zwiki/.
new patch submission process --simon, Wed, 13 Feb 2008 04:22:19 -0800 reply
Finally.. after much "fun" with the plumbing, we have an easy patch submission process. Basically, no more patch approval required. The previous process works well for many projects, but it's needlessly heavy for us right now. There was the approval delay and also, the surprising requirement to be subscribed to zwiki.org before you could darcs send a patch.
Now, just darcs send and it goes in immediately. I, and now Sascha, also have direct push/pull access. Either way, commits happen at full speed. Each new patch sends a notification to the PatchDiscussion? page for code review. For security, zwiki.org is now running on a separate, admin-maintained copy of the main repo. Also the server does not run tests for you, so do run tests yourself before sending. The CodeRepos docs have been updated accordingly.
Hacking on Zwiki code is now as easy as it should be! Here's hoping for some positive code churn.
applied patch: More functional tests: recentchanges and options --zwiki-repo, Wed, 13 Feb 2008 04:33:46 -0800 reply
- **applied patch: optimise the bottom-quoting regexp a little.. I suspect it's blowing
- up memory on mailin** --zwiki-repo, Wed, 13 Feb 2008 04:33:58 -0800 reply
applied patch: split links macro so we can omit the page-specific nav links on forms --zwiki-repo, Thu, 14 Feb 2008 12:23:05 -0800 reply
applied patch: add site navigation links to all forms --zwiki-repo, Thu, 14 Feb 2008 12:44:40 -0800 reply
... --zwiki-repo, Thu, 14 Feb 2008 12:44:55 -0800 reply
Subject: applied patch: move access key help to help page Message-ID: <20080214204440.3445278001@mail.joyful.com>
Happy Valentine's Day from Zwiki HQ --simon, Thu, 14 Feb 2008 14:24:02 -0800 reply
A happy Valentine's day to all!
Continuing with my spring cleaning theme, here are some musings.
Like many of you, I'm always wondering where is the most useful place to spend my community hacking time. Sometimes Zwiki seems like the very best place; other times it doesn't. Development pretty much came to a full stop at times last year. So I'm kicking around some ideas. First, let me get one thing off my chest: Python is a blub! There, I said it. Then there are the recurring questions of project and platform maturity, legacy cruft, how many people actually use this thing, what is the benefit, what's the future potential, etc. Don't get me wrong, I expect Python and Zope 2 to keep growing and succeeding for a very long time. And so I guess Zwiki will continue to be used and perhaps grow in user base. At the current rate of development, though, we will not stay competitive with other software or even catch up with our own bug reports. Looking at current code, I am reminded that Zwiki 0.x is mature, built on early zope 2 architecture, and that cleanup continues to be expensive (compared to starting something fresh).
Of course, many hands make light work. Imagine ten developers working on Zwiki at once - cleanup expense would no longer be noticeable, we could pretty much do whatever we want. That's not our current reality. Is it possible to find ten Zwiki developers now or in future ? I'm sure I could increase our numbers (from two active) by sustained effort, but how far and how smart a strategy that is I don't know.
Another idea which started this post is to branch, mothballing 0.x as-is and moving development focus to a Zwiki 2.x where we would do aggressive disruptive cleanup. Say we did that. What would you throw out of Zwiki 0.x ? What would you keep ?
This is to air my thoughts and also a kind of ping to gauge support for the project circa 2008, I guess. Thanks!
Happy Valentine's Day from Zwiki HQ --tralala, Fri, 15 Feb 2008 00:48:22 -0800 reply
I've been asking myself very similar questions about zwiki. I don't think python or zope are a limiting factor, however zope2 seems to encourage the coding of huge classes and zwiki has followed the fashion. I found it difficult to dig into the code and understand the concepts behind.
I believe that using zope3 or five in order to reduce the complexity and improve the readability of the code will help attracting new devs. Additionally for lowering the barrier to entry we should think of putting design documents on zwiki.org - something of a bird eye view of the project, the concepts, the assumptions, the philosophy... you get the idea.
Happy Valentine's Day from Zwiki HQ --jmax, Fri, 15 Feb 2008 09:13:20 -0800 reply
Moving away from big classes is a definite plus. I'm interested in a simpler, more extensible object model; imagine being able to easily subclass wikipage to specialize UI/behaviour for different site applications. Also, just a thought: the current comments system could be chucked out and re-done in a more OO way.
ZWiki going in earnest to Zope3 would be a motivation for me to take Zope3 seriously; unless ZWiki goes there, I'm not all that interested, as everything I do with Zope nowadays uses ZWiki as the platform.
applied patch: add nav links to contents form too --zwiki-repo, Sat, 16 Feb 2008 14:52:10 -0800 reply
- **applied patch: drop the "discussion" nav link activated by GeneralDiscussion,
- it is a special case and confusable with per-page discussion** --zwiki-repo, Sat, 16 Feb 2008 14:52:23 -0800 reply
applied patch: don't break showAccessKeys legacy url --zwiki-repo, Sat, 16 Feb 2008 14:52:24 -0800 reply
A Zwiki roadmap --Simon Michael, Sun, 17 Feb 2008 00:55:05 -0800 reply
All -
thanks for the interesting feedback on zwiki.org, zope/plone-user lists and #zwiki channel. Based on this I see the following as a good roadmap.
- Plone needs native wiki-ish solutions that fit the Plone model. It's time to stop chasing Plone, remove the CMF/Plone support (simplifing code and skins), and focus on plain Zope 2 as our platform.
- Ideally there would be one last release to fix cosmetic issues with current Plone 3 support. Anyone who wanted Zwiki in Plone would use this version. This, call it Zwiki Classic, would enter mothball/deep maintenance mode. (We might call it 1.0. Don't freak out. I'm just saying.)
- New development would focus on a 2.x branch where we would drop backwards compatibility, do very aggressive cleanup and generally make our life easier. Other priorities would be all-unicode, ease of hosting, pluggability and modularisation, moving to zope3 technologies, and extracting reusable python/zope3 libs. This would be a refactor not a ground-up rewrite - we should be able to use it pretty much right away.
- Z3wiki (the all-zope3, ZPL zwiki codebase) might be able to use and help extract libs from above. I don't expect to work on z3wiki directly myself because Zope 2 is our mainstream appserver and because I prefer to spend most of my time in GPL-land.
So there is one possible path into the future, though I don't yet know how far down it we'll go. zwiki.org has a hundred subscribers, but I'm not hearing strong support/interest/need. And there are other projects and other implementation strategies to explore. One basic issue is that with wikis proliferating and traffic intensifying, 100%-dynamic, memory-intensive wikis are not always economic. Eg for slashdottings and
for busy developers, an RCS-based solution like ikiwiki is attractive. And so on. Just pondering; we shall see. As always, further thoughts welcome.
-Simon
A Zwiki roadmap --Simon Michael, Sun, 17 Feb 2008 01:06:28 -0800 reply
PS in case you're wondering why I sent that five times - zwiki.org's iptables had started blocking incoming mail connections for some reason. Once I flushed the iptables rules, previous attempts went through.
A Zwiki roadmap --betabug, Mon, 18 Feb 2008 00:30:33 -0800 reply
Simon, I think you put it down perfectly. We'd discussed this on #zwiki, and I believe you summed it up very good!
ZWiki-unstable --simon, Fri, 22 Feb 2008 00:48:25 -0800 reply
I have pulled all the latest work from main ZWiki into the ZWiki-unicode branch, and renamed this to ZWiki-unstable. Here we'll work on the 2008 roadmap plan I mentioned (linked at FrontPage -> download box).
The next step is to set up a second zope instance running this, and a wiki. Probably a copy of zwiki.org, to give the unicode stuff a workout.
dev.zwiki.org ? unstable.zwiki.org ?
... --Michael Ang, Fri, 22 Feb 2008 00:57:50 -0800 reply
dev.zwiki.org is more generic. Any plan to move to svn?
ZWiki-unstable --simon, Fri, 22 Feb 2008 02:31:54 -0800 reply
Ok, dev.zwiki.org it is. That site (when up) is running the -unstable code.
Sorry, no plan to move to svn. Heck, I have four better alternatives. I have been talking smack about trying out git, hg or maybe bzr (in that order). I do still love darcs, even though merging those branches was darn slow.
Just now I got a little drastic and tried zwiki.org itself on -unstable, to save time. Things were worse than expected and we backpedalled. But just to warn you, I'm ready to break a few eggs:
<sm> I want to fix, remove, or retire stuff, and without a lot of expensive development time spent on trivia <sm> heck, who uses zwiki.org these days but the bots <sm>everyone who needs to has read it » sm likes this new world of unstable
zwiki.org snapshot --simon, Fri, 22 Feb 2008 03:45:44 -0800 reply
dev.zwiki.org now has a snapshot of zwiki.org content where we can debug issues.
Please don't link directly to that site, I'd like to keep it bot-free and quiet.
zwiki.org snapshot --betabug, Fri, 22 Feb 2008 23:45:48 -0800 reply
Don't forget to put in a robot.txt to exclude all bots there.
What about making the "Roadmap 2008" a proper page here, so we could also put up things like your "micro roadmap" - the one you mentioned on #zwiki for what's to do immediately:
sm notes micro-roadmap: get tests passing in ZWiki-unicode, then pull, resolve & test each patch from ZWiki
zwiki.org snapshot --Simon Michael, Sun, 24 Feb 2008 14:17:15 -0800 reply
I did so (robots.txt). RoadMap2008 ? Good idea, I was thinking the same.
update --Simon Michael, Sat, 08 Mar 2008 13:09:00 -0800 reply
Good day all, a quick update:
- I've made the zope memory monitor on zwiki.org check resident memory instead of total virtual memory. I expect this to reduce the restart frequency, which was getting silly as I ran some tests on other zope.
- I pushed a bunch of skin cleanups to ZWiki-unstable. This came from my efforts to clean up the zope 3 wiki skin and make it easier to maintain on the filesystem. It's now possible to specify an alternate filesystem skin directory (layered over zwiki) with a "skin" property (or request var.)
- I also pushed some more unicode fixes. I'm starting to feel a little more optimistic about this. There are just two major things I know that are broken - 1. indexing of some unicode content, which must be solved by newer and better catalog products, and 2. upgrading non- unicode wikis - there is some chicken and egg with updating the outline cache and unicoding the pages during upgrade. dev.zwiki.org is running the latest code and breaks on zwiki.org content, but it seems to handle the zope 3 wiki content pretty well on my macbook.
zeo --simon, Sun, 09 Mar 2008 14:56:48 -0700 reply
I have re-enabled ZEO, hoping http://zwiki.org/1358 won't bite.
unicode progress --simon, Mon, 10 Mar 2008 14:17:16 -0700 reply
The patches I've just pushed to -unstable include: "a bunch more unicode-related fixes. We can now upgradeAll the content of zwiki.org then serve it without obvious problems (complete cataloging requires unicode-aware indexes)"
This is a good milestone. If you are able, please test the dev.zwiki.org server and if you're brave, test the unstable branch on your own wikis. It does require a manual /upgradeAll to get things working. This will convert everything in your wiki from utf-8 to unicode.
The catalog situation is not bad - both Andreas' TextIndexNG3? and Dieter's ManagableIndex? are unicode aware. The latter provides a keyword index, which we need for the parents field, but I haven't got it working yet. When I do, Zwiki-unstable/unicode/2 should be usable in production..
Getting the upgrade process working - even manually - was not easy at all, because of the enmeshment of updating pages and updating the outline cache. Poor layering/encapsulation. I have been reduced to groping - testing and fixing the issues as I find them - though I do have a reasonable picture of the end goal in my head, eg the boundary line between encoded and unicode text. The unicode conversion is having the expected effect of clarifying architectural problems. Some time soon, new developer docs will emerge and help clarify things further. I'm debating whether to start these in a fresh high-signal Zwiki 2 wiki, or co-exist on zwiki.org (with cleanups).
performance fixes --Simon Michael, Tue, 11 Mar 2008 15:53:32 -0700 reply
I looked at some more server slowness today, with the help of emacs, dtach, top, vmstat, tail, highlight-lines-matching-regexp, and Tres Seaver's excellent thrashcatcher product which shows zodb load and store counts for each transaction in the trace log. I applied two fixes which large zwiki sites might want to know about:
bots were finding their way directly to page revision diffs (how ? I don't know), trying every link and causing a lot of unnecessary zodb loads. I stopped it by adding this apache rewrite rule:
# temporary to keep bots out of revisions # redirects direct requests for revisions to the latest page RewriteRule (.*)/revisions/(.*)\.[0-9]+/ $1/$2 [R,L]
visits to old revisions of zwiki.org FrontPage were generating thousands of page loads every time. The "new pages" dtml snippet on FrontPage calls pages() to show the last 5 pages, but in the revisions folder this was falling back to a non-catalog search (since there is no catalog there), loading all 2000+ old page revisions into cache. For this reason pages now returns nothing without a properly configured catalog.
- **applied patch: pages now returns nothing without a fully-configured catalog (to
- avoid blowing up =?utf-8?Q?=7Aodb?= cache when viewing old revisions)** --zwiki-repo, Tue, 11 Mar 2008 15:57:17 -0700 reply
drop outline cache ? --simon, Wed, 12 Mar 2008 15:03:51 -0700 reply
Zwiki stores the page hierarchy in two places. First, every page keeps a list of its parents, in a parents property. The hierarchy is built from this.
Second - because the above data structure is not rich enough to tell us what we want to know (what are my children ? who are my siblings ? what's the first page in the wiki ?) without thrashing the zodb - we cache the same information in the outline object in the wiki folder. This stores the whole hierarchy in one place so it's easy to query efficiently. The outline object is generated on demand and kept up to date as changes are made in the wiki (in parallel with the parents property).
I'm kicking around the idea of dropping the outline cache and keeping more detailed information on the pages - not only parents, but children and previous/next sibling. The downside of this is a lot of redundancy and (different) opportunities to get out of sync. Also you might have to touch a few more pages than before when making wiki changes. And I suppose we would have to index all these new pointers for efficient queries. The upside seems to be a simpler design that would be easier to self-repair and easier to upgrade. For example it would allow the upgrade to unicode to be handled automatically.
applied patch: make lastLog more robust, prevent occasional page breakage --zwiki-repo, Fri, 28 Mar 2008 08:10:48 -0700 reply
applied patch: make plone member email lookup more robust to fix non-mail-out (#1400) --zwiki-repo, Fri, 28 Mar 2008 08:14:18 -0700 reply
outline updating query --jmax, Mon, 07 Apr 2008 17:19:57 -0700 reply
Ironic, given the above, but...
We at Thinkubator are once again pushing against the traditional. I'm currently working in a system in which a new page can be created and "filed" under multiple parents as part of the creation step. I'm calling create() instead of edit() as it seems to offer the ability to pass a 'parents' argument. And it partially works. The parents are added locally, to the new page's property, but they don't seem to make it into the outline (and hence, to be known to the parents themselves -- imagine being in such a situation!) Now, create() seems to be calling wikiOutline().add(), which ought to do the trick, but, as I say, it isn't, and besides, when I look, I don't see an 'add()' defined anywhere for wikiObject. So... any pointers?
outline updating query --Simon Michael, Mon, 07 Apr 2008 17:40:39 -0700 reply
Hi John.. that's a little odd. Maybe you can step through with a debugger and see where it goes wrong ? Or, maybe a call to updateWikiOutline will take care of it ?
ZWiki pages incorrectly can be added in other people's Members folders --nd51, Thu, 10 Apr 2008 07:35:18 -0700 reply
- Hi,
- Can anyone shed light on which permission to set to fix this? This problem was reported by someone else on plone-users using 3.0.6. We've had this problem for some time on Plone 2.5.3. If you go to Members folder of someone other than yourself, you have permission to add a wiki page to their folder! All my permissions on security tab are set to 'manager' (with 'authenticated' for add comment) which seems to behave fine for the rest of the site.
- How do I stop this from happening without turning everything off?
- I don't want to turn off "implicitly addable" for wiki page as that'd affect the whole site. Members is an ATBTreeFolder?, as opposed to other folders being just ATFolder?. Does that make a difference?
ZWiki pages incorrectly can be added in other people's Members folders --Simon Michael, Thu, 10 Apr 2008 13:51:45 -0700 reply
Sorry.. this one does not sound familiar. If you need it fixed I am available as a consultant at joyful.com, or feel free to ask more questions on #zwiki.
- **applied patch: interim fix for the document actions display glitch with plone 3 -
- hide them** --zwiki-repo, Fri, 11 Apr 2008 15:19:16 -0700 reply
applied patch: fix the plone 3 document actions display glitch - hide them --zwiki-repo, Fri, 11 Apr 2008 15:33:55 -0700 reply
fix the plone 3 document actions display glitch - hide them --Simon Michael, Fri, 11 Apr 2008 15:37:53 -0700 reply
Quickly unpulled and recorded that one. I'll be releasing a 0.61 from the latest stable repo, which will be the one to use with plone 3.
unsubscribed an address --simon, Fri, 18 Apr 2008 13:10:16 -0700 reply
I noticed feedreader at yahoo.com subscribed to all zwiki.org edits, which seemed wrong, so I unsubscribed it. If I'm wrong, please reply.
100 subscribers --simon, Fri, 18 Apr 2008 13:15:07 -0700 reply
GeneralDiscussion has passed 100 subscribers (page + wiki). Hi!
unsubscribed an address --feedreader, Fri, 18 Apr 2008 14:39:39 -0700 reply
Simon, if my subscription seems "wrong" maybe I should think again about it... out of respect I mean. But really, it felt OK to me.
Now though it looks as if I don't need to be subscribed to notice that I've been unsubscribed, which gives me some food for thought on the subject of subscription...
- Ken
unsubscribed an address --Simon Michael, Fri, 18 Apr 2008 15:42:19 -0700 reply
Hi Ken, sorry about that. So is that your personal address ? I thought it was someone trying to pipe all edit mails into some kind of yahoo feed monitoring thing.
Neat. :) Maybe you're subscribed with another address ?
applied patch: the bones of 0.61 release notes --zwiki-repo, Tue, 22 Apr 2008 22:27:33 -0700 reply
unstable repo cleaned up! --simon, Wed, 23 Apr 2008 13:53:22 -0700 reply
I have been reviewing and consolidating branches on my local machine, zwiki.org, and wiki.zope.org, and have resolved a showstopping darcs speed issue. Pushing between the stable and unstable repos had become impractical due to the large number of conflicts I created when merging unicode and skin cleanup branches. With the help of #darcs I understood this better and re-recorded those as two conflict-free mega-patches, for a cleaner and quicker unstable repo. Also the unstable repo is now in darcs 2's backwards-compatible hashed format.
This means:
- if you have a local copy of the ZWiki-unstable repo, you should delete it and get a new copy
- you need darcs 2 to work with the unstable repo
- any version of darcs works for the stable repo
a mailing list! --simon, Wed, 23 Apr 2008 14:13:33 -0700 reply
I think I am going to set up an honest-to-god mailing list again, to make our discussions easier. Surviving on our own dogfood is good, but I want to remove that constant "where do I post this ?" friction, and also to review and discuss darcs patches easily the way other projects do (these do not travel well through wiki mail). I can set up a Zwiki team and mail list in our Launchpad project, and reuse the old zwiki gmane group. Comments ?
We'll need to figure out (anew! :) the most useful ways to integrate wiki and list, and what to do with existing page and wiki subscribers. I think think it would good to retire the GeneralDiscussion page and move all subscribers to the mail list. I suspect that would require everyone to join the Zwiki team on launchpad though. That might be a showstopper, actually.
more discussion thoughts --simon, Wed, 23 Apr 2008 21:28:26 -0700 reply
I remembered more details of the old wiki versus mail-list dilemma.. I had a 12-point memo for you, but lost it in a firefox 3 crash. Basically I have decided to try something else first: stick with the simple all-wiki setup and seek other solutions to the problems I mentioned. Eg:
- really emphasize GeneralDiscussion as the standard place for discussion, including users, developers, patch review, but excluding extremely specific topics like issues
- solve the problem of darcs patches cluttering up the page content
Getting started on 1, I archived old discussion, merged the 2008 patch discussion into GeneralDiscussion, cleaned up that page and simplified the FrontPage discussion links. Better! AboutZwikiDiscussion updates later.
who does what, new policy (was Re: New Zwiki Version?) --Simon Michael, Thu, 24 Apr 2008 09:36:48 -0700 reply
On Apr 24, 2008, at 12:44 AM, Tom Gross wrote: | Hi Simon, | | when will there be a new zwiki-version? The current one (0.60) | breaks with the recent version of PlacelessTranslationService? | (1.4.0) because most of the i18n *.po files are invalid. They lack | the "Language-code" header. | | Thanks for helping, | -Tom
Hey Tom,
thanks for the bug report, that's helpful. Also you raise an important point.
I think it's best for me to prioritise patch review, education/ coaching/mentoring, and architectural cleanup (design, code, documentation, developer and community processes), rather than churning out yet another bugfix myself. Other people need to be sending patches. So taking a page from David Roundy, whose recent "departure" from darcs has jump-started that project, I hereby announce:
I will not work on, and may not even read, any issue or feature request unless it meets one of these criteria:
- somebody sends a darcs patch, or
- somebody wants to collaborate/be mentored, or
- somebody pays me to work on it, or
- it is something that particularly affects and motivates me today
I will do a release as soon as enough blocker issues have been fixed, so if you want it to happen sooner, send darcs patches.
http://zwiki.org/DarcsRepos#how-to-use-the-main-zwiki-darcs-repository
Thanks!
two discussion helps --simon, Thu, 24 Apr 2008 10:19:37 -0700 reply
Two versions of discussion help at http://zwiki.org/GeneralDiscussion , which is better ?
two discussion helps --betabug, Thu, 24 Apr 2008 11:31:39 -0700 reply
I prefer the first one with the logic of "what do you want to do?" - "this is how you can do it".
two discussion helps --Simon Michael, Thu, 24 Apr 2008 14:11:21 -0700 reply
Agreed, thanks.
zwiki.org moved to latest code --simon, Fri, 25 Apr 2008 05:16:36 -0700 reply
I have moved zwiki.org to a fresh ZODB running the latest ZWiki-unstable code. Time to live on our own dogfood. This means zwiki.org is now all unicode - Hurrah!
Known issues: about 10% of pages are not being indexed. #1389 UnicodeDecodeError when searching has gone away.
Other changes: navigation links are available in more places (part of skins cleanup), and access keys work again.
I also doubled the ram on this VPS, from 512Mb to 1G. I haven't retuned the zopes yet, but thanks to the ram and the clean ZODB the site seems quite zippy.
zwiki.org moved to latest code --simon, Fri, 25 Apr 2008 05:18:50 -0700 reply
Correction: mailin is not broken, just upgrade teething issues. If you see errors, please briefly report them (no need for a traceback in this case.) Thanks!
applied patch: fix subscribing to the whole wiki (IssueNo0754?) --zwiki-repo, Fri, 25 Apr 2008 14:20:31 -0700 reply
zwiki.org content archived, ready for cleanup --simon, Fri, 25 Apr 2008 21:42:32 -0700 reply
old.zwiki.org is a read-only snapshot of zwiki.org as it was yesterday. (Please don't link it, to reduce server load.) I can easily zsync(*) old pages from there to the new server, should it be needed. With this in place, aggressive cleanups of zwiki.org content are safe, likely and encouraged!
(*) ZSyncer 0.7.1 was the hero of last night's migration. To my surprise, it had no problem syncing 3000 pages and random other zope objects, including the wiki catalog, outline cache, localfs objects, and so on. I had thought it would suffer all the same problems I had seen with export/import, but apparently not. Bravo!
sphinx docs --simon, Fri, 25 Apr 2008 23:35:15 -0700 reply
I'm happy to announce we now have filesystem-based docs as well as wiki ones, or rather we now have a way to publish them nicely (Sphinx, thanks to Georg Brandl).
http://zwiki.org/docs
These are generated from the restructured text .txt files in the unstable code repo, on the hour. They are mostly developer-oriented, and being close to the code, are potentially much more efficient for developers to maintain than wiki pages. (See the link for a discussion of wiki vs filesystem docs.) Also Sphinx does a nice job of pretty-printing doctests and provides a lot of support for linking and cross-referencing stuff. It does not generate api documentation from code, but hopefully we can integrate epydoc or something.
Patches welcome.
road map update --simon, Sat, 26 Apr 2008 12:52:12 -0700 reply
I have created RoadMap2008, but I think it needs an update. My current thoughts:
- as far as possible, stick with one main line of Zwiki development, rather than two. This is just simpler all round.
- the stable and unstable repos are a good idea, especially for my hosting setup; I can test newer code in production while also hosting stable sites. I imagine all unstable patches feeding into stable eventually, and all Zwiki releases coming from stable.
- keep Plone support. Refactor it, don't spend cycles supporting it, but keep the working code we have. Even partial Plone integration that just works is useful.
- refactor ourselves into a modern component-architecture application. As that is achieved, it will open more integration possibilities.
road map update --simon, Sat, 26 Apr 2008 13:03:07 -0700 reply
- within the "one main line", do mark major transitions with a version number change, eg to 1.x or 2.x. I think the switch to unicode is such a transition. I don't know exactly what we should do. I lean towards releasing one or more updates in the 0.x line, skipping 1.x entirely, and then releasing 2.0 with unicode support and some significant other features worthy of a 2.0 release. Also this should coincide with a fairly major overhaul of zwiki.org, jettisoning a lot of old content, possibly focussing it on 2.x and leaving old.zwiki.org as the 0.x support site.
Comments ?
darcs patch: Added some more functional tests (filling out the opti... --betabug.darcs, Wed, 30 Apr 2008 11:15:56 -0700 reply
darcs patch: Added some more functional tests (filling out the opti... --Simon Michael, Wed, 30 Apr 2008 11:27:41 -0700 reply
Applied, thanks!
zwiki.org mail flowing again --simon, Wed, 30 Apr 2008 11:33:01 -0700 reply
All - zwiki.org stopped sending mail after a software upgrade on april 25th. Please have a look at recent http://zwiki.org/GeneralDiscussion (starting with "zwiki.org moved to latest code --simon, Fri, 25 Apr 2008 05:16:36 -0700") to see some important announcements. Summary: zwiki.org now running latest unstable code, old.zwiki.org archived, sphinx docs, roadmap update.
applied to unstable: fix a unicode-related issue properties form breakage --zwiki-repo, Wed, 30 Apr 2008 13:56:14 -0700 reply
darcs send change --Simon Michael, Wed, 30 Apr 2008 14:20:38 -0700 reply
I've updated the description of darcs send behaviour at CodeRepos. Latest policy is this:
- darcs send to either the stable or unstable repo sends the patch to GeneralDiscussion for review. Currently this makes manual cleanup necessary on the page, but I hope to solve this soon.
- someone with direct repo access (currently Simon and Sascha) applies the patch, once approved
- there is also a secret address for each repo to which you can darcs send to bypass the review step. Ask on #zwiki for this if the review step is too much overhead for some reason.
- I myself pretty often just push patches over ssh
- however it happens, both repos send a "patch applied" notification to GeneralDiscussion when a change goes in
darcs patch: Added a simple unicode aware splitter, based on code b... --betabug.darcs, Thu, 01 May 2008 09:48:40 -0700 reply
We are using this here, so people with minimal needs don't have to install extra products. If you need more than this install a more specific splitter (e.g. a CJK splitter) create a new "ZCTextIndex Lexicon" and recreate the "SearchableText" and "Title" indexes with your new lexicon.
darcs patch: Added a simple unicode aware splitter, based on code b... --Simon Michael, Thu, 01 May 2008 10:04:42 -0700 reply
Looking good, thanks for showing the work in progress. UnicodeLexicon? seems just the missing piece for our unicode story. I like the idea that we could work with the standard product (or others like it) if installed, but that we also have a minimal version built in - at least enough to get all pages into the catalog without breaking.
applied to stable: provide indexObject to avoid revision indexing in plone 3 (#1407, Dorneles Tremea) --stable-repo, Thu, 01 May 2008 10:36:25 -0700 reply
applied to unstable: provide indexObject to avoid revision indexing in plone 3 (#1407, Dorneles Tremea) --unstable-repo, Thu, 01 May 2008 10:40:10 -0700 reply
applied to unstable: fix event log message formatting --unstable-repo, Thu, 01 May 2008 11:25:32 -0700 reply
darcs patch: Added a simple unicode aware splitter, based on code b... --betabug, Thu, 01 May 2008 11:34:59 -0700 reply
Yes, the idea is that we have something that works "out of the box" for most of the people. Of course some people will always want/need more specific stuff, so they will have to install either TextIndexNG? or something like "CJK Splitter" and recreate those indexes.
darcs patch: Added a simple unicode aware splitter, b... (and 1 more) --betabug.darcs, Thu, 01 May 2008 11:47:22 -0700 reply
applied to unstable: Added a simple unicode aware splitter, based on code by Stefan H. Holek. --unstable-repo, Thu, 01 May 2008 11:59:22 -0700 reply
applied to unstable: Create ZCTextIndexes? with Unicode splitters by default. --unstable-repo, Thu, 01 May 2008 11:59:38 -0700 reply
applied to unstable: fix LOG macro to pass tests --unstable-repo, Thu, 01 May 2008 12:47:23 -0700 reply
issuebrowser fixed --simon, Thu, 01 May 2008 20:02:46 -0700 reply
http://zwiki.org/IssueBrowser login (x, x) now works again.
applied to stable: fix excerptAt to pass its test --stable-repo, Thu, 01 May 2008 20:39:01 -0700 reply
applied to stable: fix an unsafe use of page name which could break editform (#1376, Noboyuku Koyoshi) --stable-repo, Thu, 01 May 2008 20:40:24 -0700 reply
applied to unstable: fix an unsafe use of page name which could break editform (#1376, Noboyuku Koyoshi) --unstable-repo, Thu, 01 May 2008 20:40:42 -0700 reply
applied to stable: be more careful when mutating the parents property, to avoid inconsistencies (#1353, Willi Langenberger) --stable-repo, Thu, 01 May 2008 21:02:32 -0700 reply
applied to unstable: be more careful when mutating the parents property, to avoid inconsistencies (#1353, Willi Langenberger) --unstable-repo, Thu, 01 May 2008 21:02:45 -0700 reply
Way to go! --EmmaLaurijssens, Fri, 02 May 2008 00:39:12 -0700 reply
A lot of activity again! I'm sorry I can't chime in at the moment.
applied to unstable: preserve unicode when evaluating DTML (#1411) --unstable-repo, Fri, 02 May 2008 08:55:54 -0700 reply
applied to stable: provide x access key only when external edit is available --stable-repo, Fri, 02 May 2008 11:54:58 -0700 reply
Simon Michael <simon@joyful.com> Fri May 2 11:24:12 PDT 2008
applied to unstable: provide x access key only when external edit is available --unstable-repo, Fri, 02 May 2008 12:06:01 -0700 reply
provide x access key only when external edit is available
applied to unstable: resolve sendMailTo conflict --unstable-repo, Fri, 02 May 2008 13:07:18 -0700 reply
resolve sendMailTo conflict
applied to unstable: refactor sendMailTo --unstable-repo, Fri, 02 May 2008 13:07:21 -0700 reply
refactor sendMailTo
darcs patch: Upgrade SearchableText and Title if they are TextIndex... --betabug.darcs, Fri, 02 May 2008 13:34:31 -0700 reply
- Upgrade SearchableText and Title if they are TextIndex based.
test --Simon Michael, Fri, 02 May 2008 15:00:10 -0700 reply
test
test 2 --Simon Michael, Fri, 02 May 2008 15:02:24 -0700 reply
test
test with attachment --Simon Michael, Fri, 02 May 2008 15:04:21 -0700 reply
text part
oops --simon, Fri, 02 May 2008 15:05:04 -0700 reply
Sorry, I keep forgetting to send those to TestPage.
applied to unstable: drop the special handling of darcs patch mail-ins --unstable-repo, Fri, 02 May 2008 15:07:43 -0700 reply
drop the special handling of darcs patch mail-ins
applied to unstable: forward original mail-in messages to subscribers --unstable-repo, Fri, 02 May 2008 15:07:44 -0700 reply
forward original mail-in messages to subscribers
applied to unstable: make html_quote unicode-safe, make summary and excerptAt return unicode --unstable-repo, Fri, 02 May 2008 15:07:45 -0700 reply
make html_quote unicode-safe, make summary and excerptAt return unicode
applied to unstable: fix sendMailTo --unstable-repo, Fri, 02 May 2008 16:46:37 -0700 reply
fix sendMailTo
darcs patch: Upgrade SearchableText and Title if they... (and 2 more) --betabug.darcs, Sat, 03 May 2008 04:27:31 -0700 reply
- Upgrade SearchableText and Title if they are TextIndex based.
- Minimal test for upgrading TextIndex'es through !setupCatalog.
- Extended index upgrade tests for !setupCatalog. Existing ZCTextIndexes (or anything non-TextIndex) will not be replaced. These might be there if someone set up a more advanced index than what we have by default. We're now testing for this condition. set up a different
darcs patch: Fixed the display of non-ascii search expressions sear... --betabug.darcs, Sat, 03 May 2008 04:29:20 -0700 reply
- Fixed the display of non-ascii search expressions searchwiki.
Are we drowning out any discussion? --betabug, Sat, 03 May 2008 05:50:05 -0700 reply
Sometimes when looking at the latest deluge of patches (yay! patches!) I wonder if we're drowning out any discussion that might happen on this page? Not that there was so much discussion lately.
darcs patch: Functional tests: Anonymous edits, edits... (and 1 more) --betabug.darcs, Sat, 03 May 2008 12:06:23 -0700 reply
- Functional tests: Anonymous edits, edits with options set only.
- Another functional test: expungeEditsEverywhereBy.
Are we drowning out any discussion? --Frank Laurijssens, Sun, 04 May 2008 01:13:36 -0700 reply
Well, the inclusion of the patches themselves IMHO does make the thing a bit unreadable, which makes a good discussion somewhat difficult. Did anyone notice my post earlier this week?
Are we drowning out any discussion? --betabug, Sun, 04 May 2008 03:33:05 -0700 reply
The included patches currently are getting cleaned up manually - that's why they sometimes stick around for a short while. Yeah, I noticed your post :-)
darcs patch: Make createNextIssue and changeIssueProperties unicode... --betabug.darcs, Sun, 04 May 2008 05:12:24 -0700 reply
- Make createNextIssue and changeIssueProperties unicode-safe. Did not do the same for createIssue, since comments mention this method is not used directly any more, instead it's called from createNextIssue.
Are we drowning out any discussion? --Simon Michael, Sun, 04 May 2008 10:22:12 -0700 reply
Frank, you read GeneralDiscussion on the web, rather than by email ?
Sure I noticed your post - thanks!
Are we drowning out any discussion? --Frank Laurijssens, Sun, 04 May 2008 11:05:32 -0700 reply
No, by email and on the web. I did notice you're editing out the patches but that's because I'm an all-edits (and all-wiki) subscriber. Casual subscribers to GeneralDiscussion get kilobytes of patches, to date there's no known way of editing those out :-)
And yes, I am glad development on Zwiki is catching up. I'm still in the process of rebuilding my home server, which is taking all the spare time I have at the moment. Virtualization is great except when your storage subsystem fails and your backups only contain the essential data. The good thing about this is that I needed to look into a couple of things anyway, so this is a good training experience and I'm glad I got a round tuit.
developer chat summary --simon, Sun, 04 May 2008 12:06:20 -0700 reply
Good morning.. Sascha and I just had one of our impromptu developer conferences on #zwiki. We chatted for about an hour, brainstorming priorities and the best ways to go forward.
Here are some topics and decisions that came up:
- sm points to great motivating talks on software quality, collaboration, documentation: http://www.oopsla.org/oopsla2007/index.php?page=podcasts/ , David Parnas and Fred Brooks
- sm really, really would like to have the best of both wiki pages and revision-controlled files, but needs to live with current darcs/wiki docs split for now. I think that means more sphinx (fs-based) docs, especially developer/testing oriented docs
- brainstorming a future storage-agnostic static-html-generating easy-hosting componentised zwiki
- betabug just wants to smoke out a lot of bugs and have a proper unicode-savvy zwiki. sm agrees with that goal as next step - a good bug-free unicode zwiki, that does the things we're familiar with
- we agree on really easy filesystem-based skinning as a valuable thing to tackle after that
- sm is looking for ways we can simplify to focus our energy on new stuff
- stick with the plan of freezing utf branch after one more bugfix release, and focussing all energy on unicode branch
- are we sure the unicode retrofitting can become trustably robust for all users with reasonable effort ? We think so. To get there, we need a clear specification document, comprehensive testing, and more code review. We should be ready to change zwiki design if necessary to make this easier.
- no darcs 2 for osx ppc makes it hard for betabug to work on zwiki at the office
- efficient discussion and patch review/submission processes remain a top priority
And some definite actions:
- sm to keep streamlining patch submission process, eg finish mailin enhancement to deliver darcs patches without posting them on the wiki
- sm to release a bugfixed utf as 0.61 - this needs me to put out remaining fires with plone 3.x support
- betabug to focus on unicode testing
- sm to move my latest unicode notes to a sphinx doc, which we should develop into a specification to guide testing and cleanup
- sm and betabug to start doing joint code review sessions
Are we drowning out any discussion? --Simon Michael, Sun, 04 May 2008 12:26:32 -0700 reply
> No, by email and on the web. I did notice you're editing out the > patches but that's because I'm an all-edits (and all-wiki) > subscriber. Casual subscribers to GeneralDiscussion get kilobytes of > patches, to date there's no known way of editing those out :-)
I see. Well the plan (the next step) as I see it is to have it keep delivering the full patch emails to subscribers, but strip out the code before posting to the wiki page. So yes, casual subscribers will still see the patches. Is that really a problem in a mailreader, where you can easily skip them ? Developers need to see the code for patch review, casual subscribers can benefit from seeing activity, and I feel the advantage of having a single discussion area for everyone is large (at the moment). If more of you have an opinion, please send it.
- **applied to unstable: make displaysSubtopicsWithDtml regexp a little
- smarter** --unstable-repo, Thu, 08 May 2008 16:12:25 -0700 reply
make displaysSubtopicsWithDtml regexp a little smarter
Simon Michael <simon@joyful.com>
- **applied to unstable: add our proposed function/method contract
- documentation guidelines to the style guide** --unstable-repo, Sat, 10 May 2008 11:57:56 -0700 reply
add our proposed function/method contract documentation guidelines to the style guide
Simon Michael <simon@joyful.com>
applied to unstable: docstring --unstable-repo, Sat, 10 May 2008 11:58:06 -0700 reply
docstring
Simon Michael <simon@joyful.com>
applied to unstable: some text change --unstable-repo, Sat, 10 May 2008 11:58:08 -0700 reply
some text change
Simon Michael <simon@joyful.com>
- **applied to unstable: update the copyright message to reflect de facto
- status.. I thought I did this long ago** --unstable-repo, Sat, 10 May 2008 11:58:10 -0700 reply
update the copyright message to reflect de facto status.. I thought I did this long ago
Simon Michael <simon@joyful.com>
applied to unstable: contract docs cleanups --unstable-repo, Sat, 10 May 2008 12:07:27 -0700 reply
contract docs cleanups
Simon Michael <simon@joyful.com>
clarifying state dependencies --Simon Michael, Sat, 10 May 2008 12:18:55 -0700 reply
A couple of days away from keyboard are good for thinking. After another productive chat with betabug this morning, I have added some new guidelines to the developer style guide, which I include here for your interest and critique.
Function/method contracts
We are considering ways to document function and method "contracts". Here's an example. We are most interested in
their input (argument) and output (return value) types
their degree of statefulness/statelessness. Are they
- pure functions depending only on the arguments ?
- well-behaved methods which depend on/modify only self ?
- or do they depend on/modify other pages, the wiki folder, other things ?
Understanding statefulness in particular should help us to identify and gather together the most stateful code, reduce state dependencies, and increase the proportion of pure and semi-pure functional code, which will greatly aid testing, debugging and reliability.
Here is a possible convention for now: when you touch code, review the docstring and try to add a comment to the function definition describing (a) the return type(s), (b) state that it depends on other than the arguments, if any, and (c) state that it modifies other than the return value, if any. Even a rough guess at the state dependencies could be helpful. Here are some examples. The dependency on self for permission checking is omitted.:
def method(args) # -> return value type(s) [; depends on: ...] [; modifies: ...] def htmlquote(self, text): # -> string def asAgeString(self,time): # -> string | empty string ; depends on: self (for current time) def excerptAt(self, expr, size=100, highlight=1, text=None): # -> html string | empty string ; depends on: self (if no text provided) def revisionNumberBefore(self, username): # -> revision number | none ; depends on: self, revisions def handleSubtopicsProperty(self,subtopics,REQUEST=None): # -> none ; modifies: self def expungeEditsEverywhereBy(self, username, REQUEST=None, batch=0): # -> none ; depends on: all pages, revisions ; modifies: all pages, revisions def upgradeAll(self,render=1,batch=0,REQUEST=None): # -> none; depends on: wiki; modifies: wiki (folder, pages, dtml methods, catalog, outline, revisions..)
applied to unstable: begin adding method contract annotations --unstable-repo, Sat, 10 May 2008 12:45:03 -0700 reply
begin adding method contract annotations
Simon Michael <simon@joyful.com>
applied to unstable: more doc cleanups --unstable-repo, Sat, 10 May 2008 12:45:06 -0700 reply
more doc cleanups
Simon Michael <simon@joyful.com>
- **applied to unstable: move the mailin external method to a normal page
- method** --unstable-repo, Sat, 10 May 2008 20:02:32 -0700 reply
move the mailin external method to a normal page method Incoming mail should now be posted to SOMEPAGEURL/mailin. This means you need to know a valid page id in /etc/aliases (FrontPage or ZWiki are good choices), but removes one more configuration hassle.
- **applied to unstable: =?utf-8?Q?initiali=7Ae?= mail mixin classes to
- ensure permission enforcement.. still finding these =?utf-8?Q?=3F?=** --unstable-repo, Sat, 10 May 2008 20:02:53 -0700 reply
initialize mail mixin classes to ensure permission enforcement.. still finding these ?
- **applied to unstable: dagnabbit.. fix double headers in mail-throughs
- due to premature code** --unstable-repo, Sat, 10 May 2008 20:11:42 -0700 reply
dagnabbit.. fix double headers in mail-throughs due to premature code
applied to unstable: more contract documenting tips --unstable-repo, Sat, 10 May 2008 22:09:00 -0700 reply
more contract documenting tips
applied to unstable: document remaining Mail.py contracts --unstable-repo, Sat, 10 May 2008 22:09:12 -0700 reply
document remaining Mail.py contracts
applied to unstable: mailin refactoring, cleanup, contract documenting --unstable-repo, Sat, 10 May 2008 22:09:14 -0700 reply
mailin refactoring, cleanup, contract documenting
- **applied to unstable: Upgrade SearchableText? and Title if they are
- TextIndex? based.** --unstable-repo, Sun, 11 May 2008 11:54:10 -0700 reply
Upgrade SearchableText? and Title if they are TextIndex? based.
- **applied to unstable: Minimal test for upgrading TextIndex?'es through
- setupCatalog.** --unstable-repo, Sun, 11 May 2008 11:54:27 -0700 reply
Minimal test for upgrading TextIndex?'es through setupCatalog.
applied to unstable: Extended index upgrade tests for setupCatalog. --unstable-repo, Sun, 11 May 2008 11:54:30 -0700 reply
Extended index upgrade tests for setupCatalog. Existing ZCTextIndexes? (or anything non-TextIndex?) will not be replaced. These might be there if someone set up a more advanced index than what we have by default. We're now testing for this condition. set up a different
- **applied to unstable: Functional tests: Anonymous edits,
- edits with options set only.** --unstable-repo, Sun, 11 May 2008 12:32:50 -0700 reply
Functional tests: Anonymous edits, edits with options set only.
- **applied to unstable: Another functional test:
- expungeEditsEverywhereBy.** --unstable-repo, Sun, 11 May 2008 12:33:07 -0700 reply
Another functional test: expungeEditsEverywhereBy.
- **applied to unstable: Make createNextIssue and changeIssueProperties
- unicode-safe.** --unstable-repo, Sun, 11 May 2008 12:33:08 -0700 reply
Make createNextIssue and changeIssueProperties unicode-safe. Did not do the same for createIssue, since comments mention this method is not used directly any more, instead it's called from createNextIssue.
- **applied to unstable: Fixed the display of non-ascii search
- expressions searchwiki.** --unstable-repo, Sun, 11 May 2008 12:33:16 -0700 reply
Fixed the display of non-ascii search expressions searchwiki.
applied to unstable: drop the unused spam reporting mailin feature --unstable-repo, Sun, 11 May 2008 19:01:20 -0700 reply
drop the unused spam reporting mailin feature
applied to unstable: clean up mail code and tests --unstable-repo, Sun, 11 May 2008 19:01:21 -0700 reply
clean up mail code and tests
oops --simon, Sun, 11 May 2008 19:02:23 -0700 reply
Sorry for the resend of those last four notifications.. darcs pull --dry-run seems to trigger them. betabug I verified all your latest patches are in ZWiki-unstable.
great --Simon Michael, Sun, 11 May 2008 19:13:58 -0700 reply
PS, thanks for the unicode work and all the tests! These obviously help a lot.
- **applied to stable: make displaysSubtopicsWithDtml regexp a little
- smarter** --stable-repo, Sun, 11 May 2008 19:15:59 -0700 reply
make displaysSubtopicsWithDtml regexp a little smarter
darcs patch: 1171 - more than one double parenthesis link in the sa... --betabug.darcs, Wed, 14 May 2008 07:37:41 -0700 reply
- Wed May 14 17:32:09 EEST 2008 betabug.darcs@betabug.ch
- 1171 - more than one double parenthesis link in the same paragraph works now.
- **applied to unstable: 1171 - more than one double parenthesis link in
- the same paragraph works now.** --unstable-repo, Wed, 14 May 2008 12:12:33 -0700 reply
1171 - more than one double parenthesis link in the same paragraph works now.
applied to unstable: make fixperms fix the top directory also --unstable-repo, Wed, 14 May 2008 12:24:24 -0700 reply
make fixperms fix the top directory also
applied to stable: make fixperms fix the top directory also --stable-repo, Wed, 14 May 2008 12:30:19 -0700 reply
make fixperms fix the top directory also
darcs patch: 356 - the asterisk "*" is an allowed character in URLs?. --betabug.darcs, Thu, 15 May 2008 07:18:21 -0700 reply
- Thu May 15 17:05:20 EEST 2008 betabug.darcs@betabug.ch
- 356 - the asterisk "*" is an allowed character in URLs?.
darcs patch: 1361 - determine permissions on editform() at runtime. --betabug.darcs, Thu, 15 May 2008 07:19:23 -0700 reply
- Thu May 15 16:34:03 EEST 2008 betabug.darcs@betabug.ch
- 1361 - determine permissions on editform() at runtime.
The editform() form should be displayed also in the case that the user has "Add" rights on the wiki, but no "Edit" rights on the page he started from. (E.g. going through the "add" form on a restricted page, then clicking on the "Preview" button raised Unauthorized before this fix.)
darcs patch: 1414 - pre-line for "pre" blocks is not ... (and 1 more) --betabug.darcs, Fri, 16 May 2008 00:46:09 -0700 reply
- Fri May 16 10:24:04 EEST 2008 betabug.darcs@betabug.ch
- 1414 - pre-line for "pre" blocks is not what we want, it kills indentation.
- Fri May 16 10:30:10 EEST 2008 betabug.darcs@betabug.ch
- Fixed some minor incorrect css.
- **applied to unstable: 1361 - determine permissions on editform() at
- runtime.** --unstable-repo, Fri, 16 May 2008 07:23:17 -0700 reply
1361 - determine permissions on editform() at runtime. The editform() form should be displayed also in the case that the user has "Add" rights on the wiki, but no "Edit" rights on the page he started from. (E.g. going through the "add" form on a restricted page, then clicking on the "Preview" button raised Unauthorized before this fix.)
- **applied to unstable: 356 - the asterisk "*" is an allowed character
- in URLs?.** --unstable-repo, Fri, 16 May 2008 07:23:28 -0700 reply
356 - the asterisk "*" is an allowed character in URLs?.
applied to unstable: Fixed some minor incorrect css. --unstable-repo, Fri, 16 May 2008 07:23:30 -0700 reply
Fixed some minor incorrect css.
- **applied to unstable: 1414 - pre-line for "pre" blocks is not what we
- want, it kills indentation.** --unstable-repo, Fri, 16 May 2008 07:23:33 -0700 reply
1414 - pre-line for "pre" blocks is not what we want, it kills indentation.
applied to unstable: clean up test rules --unstable-repo, Sat, 17 May 2008 10:16:21 -0700 reply
clean up test rules
applied to unstable: more mail test cleanups --unstable-repo, Sat, 17 May 2008 10:16:40 -0700 reply
more mail test cleanups
applied to unstable: clean up test rules --Simon Michael, Sat, 17 May 2008 10:18:00 -0700 reply
Sascha, any idea why (with this latest makefile cleanup) make testmod_Functional or make testmod_Functional_test breaks ? It runs fine during make test.
applied to unstable: remove some old make rules --unstable-repo, Sat, 17 May 2008 11:05:25 -0700 reply
remove some old make rules
applied to unstable: move around some mail tests --unstable-repo, Sat, 17 May 2008 11:05:31 -0700 reply
move around some mail tests
- **applied to unstable: make test_methodregexp no longer has to match
- from the beginning** --unstable-repo, Sat, 17 May 2008 11:05:34 -0700 reply
make test_methodregexp no longer has to match from the beginning
- **applied to unstable: slight stx tests cleanup and a failing test for
- #1115** --unstable-repo, Sat, 17 May 2008 11:59:55 -0700 reply
slight stx tests cleanup and a failing test for #1115
applied to unstable: make docs --unstable-repo, Sat, 17 May 2008 12:00:03 -0700 reply
make docs
applied to unstable: style docs --unstable-repo, Sat, 17 May 2008 12:00:05 -0700 reply
style docs
applied to unstable: contract docs --unstable-repo, Sat, 17 May 2008 12:00:12 -0700 reply
contract docs
applied to unstable: rename and add a test for obfuscateEmailAddresses --unstable-repo, Sat, 17 May 2008 12:14:07 -0700 reply
rename and add a test for obfuscateEmailAddresses
darcs patch: 939 - suppress "e" accesskey in editform. --betabug.darcs, Sat, 17 May 2008 12:14:32 -0700 reply
- Sat May 17 21:58:01 EEST 2008 betabug.darcs@betabug.ch
- 939 - suppress "e" accesskey in editform.
applied to unstable: 939 - suppress "e" accesskey in editform. --unstable-repo, Sat, 17 May 2008 12:21:32 -0700 reply
939 - suppress "e" accesskey in editform.
darcsweb urls --Simon Michael, Sat, 17 May 2008 12:31:18 -0700 reply
The following guessable urls now take you to the darcsweb index:
http://zwiki.org/darcsweb http://joyful.com/darcsweb http://joyful.com/darcs
Markup hiding --frank, Sun, 18 May 2008 20:41:34 -0700 reply
I closed #1001, because I think that if we want to implement this properly, we might as well get rid of page types entirely. I mean, using RST, STX, HTML or any other type is just the preference of the user when editing a page. Why bother other users upon subsequents edits with the page type the creator of the page chose? IMHO, this is a Zope issue, not a Zwiki issue.
cia.vc --betabug, Fri, 30 May 2008 02:14:18 -0700 reply
A while back sm and me talked about getting cia.vc connected to zwiki CodeRepos again. Apparently the solution given from cia.vc is a kludge (you have to replace the darcs binary with a wrapper script). Today looking at the darcs page I noticed that perhaps it could be possible to use darcs-monitor for this? It's made to send out mails e.g. from a posthook, and judging from the readme, it lets you create a template.
applied to unstable: 517 - refactor updating backlinks. --unstable-repo, Sat, 31 May 2008 02:09:34 -0700 reply
517 - refactor updating backlinks. This will now do all kind of nice stuff, like respect the link style that the link on the page previously had. Added a bunch of tests too.
- **applied to unstable: 517 - updating backlinks does not create links
- out of plain text any more.** --unstable-repo, Sat, 31 May 2008 02:09:51 -0700 reply
517 - updating backlinks does not create links out of plain text any more. Problem existed when the canonical ID of the old page matched some plain text word (which happens only when the word in question is not a WikiWord?). We're now checking this condition. But there are other problems with this method: It does only bracket freeform links and replaces sometimes with bracket links what could be WikiLinks? - it will also not gracefully handle other styles of freeform links (double-bracket or double-parenthesis). A more thorough patch is in the works, this is here so there is a minimal working version to compare.
darcs patch: ZTinyMCE? integration and setup. --betabug.darcs, Thu, 12 Jun 2008 13:18:12 -0700 reply
- Thu Jun 12 23:08:01 EEST 2008 betabug.darcs@betabug.ch
- ZTinyMCE integration and setup.
When peterbe's ZTinyMCE product is installed and configured, pagetypes that "support Epoz" will pick up TinyMCE as their "Rich Text Editor" (which works rather well, even in Safari 3). There is also a AnyPage/setupTinyMCE method, which will set up the ZTinyMCE instances in the wiki folder, with a configuration that works (some configuration options have a problem with our standard error message).
Patch is currently in my repo.
- **applied patch: make plone member email lookup more robust to fix
- non-mail-out (#1400)** --ben jack, Sat, 21 Jun 2008 15:22:24 -0700 reply
...
Help displaying hierarchy --jmax, Thu, 03 Jul 2008 14:45:12 -0700 reply
I'm trying to insert the whole contents/hierarchy listing in a sidebar. And failing. The contentspage simply calls "options/hierarchy" but I'm unable to get this to work by calling it myself (I keep getting a keyerror for 'hierarchy'). What does "options" mean in this context, and where is the hierarchy rendering code anyway? Any help appreciated.
Help displaying hierarchy --Simon Michael, Thu, 03 Jul 2008 17:58:42 -0700 reply
John, I believe the contentspage method in View.py passes hierarchy as an argument to contentspage.pt, accessed via "options". Have a look there to see what it calls to render that. (I'm a bit off-net, getting a hard drive repaired.)
Help displaying hierarchy --betabug, Thu, 03 Jul 2008 22:17:19 -0700 reply
The relevant code is in OutlineSupport?.py, in the contents() method (which returns the contentspage() method sm mentioned). What later becomes "hierarchy" is set up as self.renderNesting(combos,here=here) there. But then, contents() goes some way to build up "combos", so you can't just leave it out I think. See if you can use context/contents in you Page Template.
- **applied to unstable: add "make summary" rule to generate a code
- overview** --unstable-repo, Thu, 10 Jul 2008 04:08:38 -0700 reply
add "make summary" rule to generate a code overview
applied to unstable: make summaryhtml --unstable-repo, Tue, 05 Aug 2008 17:06:07 -0700 reply
make summaryhtml
applied to unstable: update copyright date --unstable-repo, Tue, 05 Aug 2008 17:17:56 -0700 reply
update copyright date
po file not working --geon, Thu, 28 Aug 2008 10:07:49 -0700 reply
We put cs.po into directory i18n, restart zope, in conf there is lang cs, and still we do not see czech translations. What do we wrong? Thank you.
po file not working --geon, Mon, 01 Sep 2008 00:10:55 -0700 reply
Noone can help? Or this is not the best place where to ask such a questions? You know, we would like to make our web available even for those who can not speak english well enough .... I translated nearly all the terms and after we try it, it could append to official distribution of Zwiki.
po file not working --betabug, Tue, 02 Sep 2008 00:21:42 -0700 reply
Geon, sorry for the long delay. Do you get any error messages in the Zope error log or when you start Zope with ./zopectl fg? I'm myself not really familiar with the translation machinery (and sm seems to be busy right now), but I'll promise to have a look if you can put the traceback somewhere.
po file not working --geon, Tue, 02 Sep 2008 05:36:08 -0700 reply
http://geon.wz.cz/sharing/zwikilog1.log
site outage --simon, Fri, 05 Sep 2008 12:34:59 -0700 reply
I upgraded to ubuntu hardy this morning, and became so overconfident I forgot to check sites (!). All sites were down briefly, zwiki was down for longer since I had forgotten some config details. All should be back to normal now.
po file not working --Simon Michael, Fri, 05 Sep 2008 12:48:43 -0700 reply
Hi geon.. this is definitely the right place to ask (there's also #zwiki on irc.freenode.net). Your translation will be a welcome addition. Your log shows that PTS doesn't like the po file yet, you need to fix the things it's complaining about. Compare with one of the other templates to see what's needed.
site outage --Simon Michael, Fri, 05 Sep 2008 12:50:41 -0700 reply
PS, please report issues if you find them. I see darcsweb has broken.
bug in search? --Simon Michael, Fri, 05 Sep 2008 13:00:49 -0700 reply
This is probably the most frequent error we see, and it's a complex topic. I don't remember all the details but I will. Here are some tips:
0. search the release notes and i18n pages on zwiki.org for background info and how-tos
1. if you have a page containing non-UTF-8 characters, you should fix that. http://www.py.cz/FrontPage/fixEncoding or http://www.py.cz/FrontPage/upgradeAll
may take care of that for you. Watch the event log at BLATHER
priority if you want to know what it's doing.
2. the ZWiki-unstable branch (http://zwiki.org/DarcsRepos) has native unicode support which may or work better for you. This is the way of the future, it's what zwiki.org uses so it's worth a try (on a copy of your wiki, in case it doesn't work out.) The release notes describe how to convert the wiki to unicode. You might need to reconfigure your zcatalog index as well.
Congratulations on your site by the way, it looks good. Nice to see some new Zwiki sites.
site outage --Simon Michael, Fri, 05 Sep 2008 13:01:19 -0700 reply
> PS, please report issues if you find them. I see darcsweb has broken.
Fixed.
tags for Python`s versions --geon, Wed, 17 Sep 2008 18:30:00 -0700 reply
Python 3000 is coming... On our web there are lots of code examples valid for 2.5. It is sure that new examples could be valid for either py25, py26 or py3000. You, with you wikis, might have the same trouble: how to nicely "tag" codes or even whole page that they are valid for relevant version of Python? As for me I would like to see something like: http://geon.wz.cz/sharing/tagsforwiki.png. Do you have any idea ho to make it? Do you have any other idea? Thank you
tags for Python`s versions --betabug, Thu, 18 Sep 2008 01:11:25 -0700 reply
Basically what you want to do is to put a "class" on the "literal block" (in !reStructuredText) or "pre" block (in html). Then you can use CSS to display the python version number accordingly. I don't know how to do this in RST though.
tags for Python`s versions --betabug, Thu, 18 Sep 2008 01:16:09 -0700 reply
The class attribute could likely do it. With something like:
.. class:: py25 :: print "python example here"
Which would result in:
print "python example here"
tags for Python`s versions --geon, Thu, 18 Sep 2008 01:32:41 -0700 reply
Well I always thought that that css can only change colour, height, etc. and not add any text to source. If I am wrong, pls correct me. I have played around with Admonition and that could be the way.
tags for Python`s versions --betabug, Thu, 18 Sep 2008 01:53:20 -0700 reply
CSS can do much more than just change colour, height etc. for example it can display a background image - which probably is what you want to do to have things appear like in your example. In theory you could even add some text, but I think support in older browsers is a bit lacking.
tags for Python`s versions --geon, Thu, 18 Sep 2008 08:22:52 -0700 reply
Yes, that might be it, except for, that the image must be very transparent not to cover possible text. Anyway, all admonitions can all be one-liners except for admonition-and-by-the-way. Is it possible to make it one-liner too? It would perfectly fit to my case. I could just use .. admonition:: Python 3000, do a correct css, and that's all.
tags for Python`s versions --geon, Thu, 18 Sep 2008 11:55:34 -0700 reply
So finally there are two results, it might help someone:
1/ img on a background:
.. class:: py25
belonging css:
.py25 {
background-image: url("py25.png");
background-position: right top;
background-repeat: no-repeat;}
disadvantage: sometimes there is a text over it
2/ admonition - div moved to the right
.. admonition:: Python 3.0
belonging css:
div.admonition-python{
background-color:#ffe8c4;
margin-left: 1em;
border: medium outset;
text-align:center;
padding: 1em;
width:80px;
float: right;
clear: right;}
disadvantage: slightly more difficult, less intuitive. But the text perfectly floats around it.
- **applied to unstable: expose renderText so I can format markups other
- than stx from dtml** --unstable-repo, Fri, 17 Oct 2008 15:24:03 -0700 reply
expose renderText so I can format markups other than stx from dtml Ignore-this: 4b59ac88b550ae2651b291d9be6642cc
- **applied to stable: expose renderText so I can format markups other
- than stx from dtml** --stable-repo, Fri, 17 Oct 2008 15:26:08 -0700 reply
expose renderText so I can format markups other than stx from dtml Ignore-this: 4b59ac88b550ae2651b291d9be6642cc
applied to stable: release note --stable-repo, Fri, 24 Oct 2008 22:23:54 -0700 reply
release note Ignore-this: 8bb013316280cc661e6fcb36dd93a8e4
- **applied to stable: update makefile checklists,
- note new CHANGES.txt suffix** --stable-repo, Fri, 24 Oct 2008 22:24:28 -0700 reply
update makefile checklists, note new CHANGES.txt suffix Ignore-this: 4a383bf787ff099c4edc915c0af126a6
applied to stable: bump version to 0.61.0 --stable-repo, Fri, 24 Oct 2008 22:24:33 -0700 reply
bump version to 0.61.0 Ignore-this: b2f5a904942137847f157d7d1fc4a198
applied to stable: TAG release-0-61-0 --stable-repo, Fri, 24 Oct 2008 22:24:40 -0700 reply
TAG release-0-61-0 Ignore-this: 4c4198a23b7360760a16d8a9e43e2540
applied to stable: updating the pot file.. fingers crossed --stable-repo, Fri, 24 Oct 2008 22:25:33 -0700 reply
updating the pot file.. fingers crossed Ignore-this: c88a224320f8c16016e4170c35531f4b
applied to unstable: release note --unstable-repo, Fri, 24 Oct 2008 22:37:51 -0700 reply
release note Ignore-this: 8bb013316280cc661e6fcb36dd93a8e4
- **applied to unstable: update makefile checklists,
- note new CHANGES.txt suffix** --unstable-repo, Fri, 24 Oct 2008 22:37:55 -0700 reply
update makefile checklists, note new CHANGES.txt suffix Ignore-this: 4a383bf787ff099c4edc915c0af126a6
applied to unstable: updating the pot file.. fingers crossed --unstable-repo, Fri, 24 Oct 2008 22:37:59 -0700 reply
updating the pot file.. fingers crossed Ignore-this: c88a224320f8c16016e4170c35531f4b
applied to unstable: bump version to 0.61.0 --unstable-repo, Fri, 24 Oct 2008 22:38:02 -0700 reply
bump version to 0.61.0 Ignore-this: b2f5a904942137847f157d7d1fc4a198
0.61 released, new repos --Simon Michael, Fri, 24 Oct 2008 23:38:00 -0700 reply
Good morning all,
I have just released the stable branch as Zwiki 0.61, as part of catch- up and cleanup. You can see the release notes at http://zwiki.org/docs/CHANGES.html . This gets the Plone 3.x layout fix into the released version.
Also I have converted both stable and unstable repos to darcs-2 format. This means you now need darcs 2.x to talk to the zwiki.org repos. Also your local zwiki repos are now out of date and you must get a fresh copy from zwiki.org. Sorry about that, but it had to be done. See http://zwiki.org/DarcsRepos for darcs download link and the zwiki repo info.
Best, -Simon
ZWiki 0.61 tarball fixed --Simon Michael, Tue, 28 Oct 2008 15:17:38 -0700 reply
tessier alerted me that http://zwiki.org/releases/ZWiki-0.61.0.tgz had bad paths in it - now fixed. I have finally acted on an old makefile todo and tarballs now contain exactly what you get by doing darcs get. Thanks!
FrontPage cleaned up --simon, Thu, 11 Dec 2008 17:11:37 -0800 reply
A cleanup burst after a zwiki-related chat on irc.
I'll be in Ireland for two weeks. Happy holidays all!
FrontPage cleaned up --EmmaLaurijssens, Fri, 12 Dec 2008 04:18:12 -0800 reply
Happy holidays, Simon! I noticed that on IE6, the sidebar is not really a sidebar, the main content starts below the sidebar.
FrontPage cleaned up --EmmaLaurijssens, Fri, 12 Dec 2008 04:23:33 -0800 reply
Fixed by decreasing the main width from 80% to 77%. Does this have any unwanted side effects in other browsers?
darcs patch: Added some functional tests. (and 1 more) --Simon Michael, Sat, 19 Jan 2008 12:20:11 -0800 reply
Yay, tests! Applied.
server trouble --simon, Sat, 19 Jan 2008 12:47:12 -0800 reply
The server (and all sites) has been restarting much more frequently since the middle of last week. I suspect some new bot traffic. I raised the memory limit a little and am working on the move to a new server.
sites moved --Simon Michael, Mon, 21 Jan 2008 18:48:20 +0000 reply
Well, this was the bumpiest upgrade ever.
Sites are now finally running on the new server, after a day or two of downtime. I'm very sorry about the outage to the wikis. I assumed this move would be straightforward, didn't realize I was dealing with a new apache version, rushed it, got busy offline, and had to dig quite hard to find what was causing sites to hang after a couple of requests (a combination of apache and zope config issues).
I now have 512M of ram to play with (up from 256) and a modern ubuntu gutsy system. All sites should be running better than before, aside from a few remaining things being resolved (darcsweb). Other problem reports are welcome. I'll be doing more tuning as needed.
So, basically, good news! Thanks and happy new year to all!