Re: [Twisted-web] Status of divmod-dev list?
On 15 Dec, 04:28 pm, werner@thieprojects.ch wrote:
Hi all
coming back to this thread to check if by any chance it was possible to find the culprit for the disappearing messages.
Of course all the time is sucked into the Xmas vortex...
The server hosting divmod.org is now more or less defunct. The content will reappear at some point, but I'm not ready to predict exactly where or when. Hopefully the divmod-dev will re-appear at some point, but not necessarily in its previous form (it may be a launchpad mailing list, for example).
Jean-Paul
On 12/19/10 5:32 PM, exarkun@twistedmatrix.com wrote:
On 15 Dec, 04:28 pm, werner@thieprojects.ch wrote:
Hi all
coming back to this thread to check if by any chance it was possible to find the culprit for the disappearing messages.
Of course all the time is sucked into the Xmas vortex...
The server hosting divmod.org is now more or less defunct. The content will reappear at some point, but I'm not ready to predict exactly where or when. Hopefully the divmod-dev will re-appear at some point, but not necessarily in its previous form (it may be a launchpad mailing list, for example).
Jean-Paul
Twisted-web mailing list Twisted-web@twistedmatrix.com http://twistedmatrix.com/cgi-bin/mailman/listinfo/twisted-web
Fine with me as long as nevow stays with us...
Werner
I'd like to add that I'm a bit more concerned about divmod.org accessibility, especially now that I've got a project that depends on Nevow.
I put a tarball of Nevow-0.10.0+r18026.tar.gz on my download site for now, but I'm glad to do anything I can to help get the repo back up.
If there's no interest in hosting the repo, lets move it to launchpad or github or something. If those concerned have no time, I'm glad to steward the Nevow source for awhile.
-phil
On Dec 19, 2010, at 11:32 AM, exarkun@twistedmatrix.com wrote:
On 15 Dec, 04:28 pm, werner@thieprojects.ch wrote:
Hi all
coming back to this thread to check if by any chance it was possible to find the culprit for the disappearing messages.
Of course all the time is sucked into the Xmas vortex...
The server hosting divmod.org is now more or less defunct. The content will reappear at some point, but I'm not ready to predict exactly where or when. Hopefully the divmod-dev will re-appear at some point, but not necessarily in its previous form (it may be a launchpad mailing list, for example).
Jean-Paul
Twisted-web mailing list Twisted-web@twistedmatrix.com http://twistedmatrix.com/cgi-bin/mailman/listinfo/twisted-web
On Sun, Dec 19, 2010 at 7:52 PM, Phil Christensen phil@bubblehouse.orgwrote:
I'd like to add that I'm a bit more concerned about divmod.orgaccessibility, especially now that I've got a project that depends on Nevow.
Is it not yet time that Nevow replaces Woven and gets adopted into Twisted?
On Dec 19, 2010, at 12:56 PM, Colin Alston wrote:
On Sun, Dec 19, 2010 at 7:52 PM, Phil Christensen phil@bubblehouse.org wrote: I'd like to add that I'm a bit more concerned about divmod.org accessibility, especially now that I've got a project that depends on Nevow.
Is it not yet time that Nevow replaces Woven and gets adopted into Twisted?
This will involve some work, and so we would not want to wait for this to happen in order to make the code accessible again.
Also, not all of Nevow will be moved into Twisted. Currently the plan (as defined by some conversations between Jean-Paul and myself) is to move the resource model, and the templating layer (omitting all the deprecated crud with the 'context' object from both), in two separate steps. We're not planning to move Athena into Twisted, as it's a large pile of aging JavaScript, and would involve a significant amount of browser-based testing in order to get right, which Twisted doesn't currently need. Twisted wants to generate HTML in a few places - directory listings, error pages - but doesn't currently need any two-way browser communication, and even if it did, Athena would need to be stripped down to a much smaller core in order to be appropriate for this.
If you'd like to help with that then please let me know. As you can see from the steadily-increasing open ticket count in the weekly emails, we need more people to help in general.
On 01:15 am, foom@fuhm.net wrote:
On Dec 19, 2010, at 7:25 PM, Glyph Lefkowitz wrote:
Currently the plan is to move the resource model [..]
You're kidding, right? :(
I think "resource model" is a pretty vague way to explain the plan (the next phase of the plan is actually to identify specifically (and _write down_) what should be copied into Twisted Web). Here are some specific things that Twisted Web should adopt from Nevow:
- locateChild - Deferreds in Resource lookup - Deferreds in response rendering - Something like `NotFound`
Probably some other things too. Are these the things about "resource model" that make you sad? Or did the phrase conjure something else?
Jean-Paul
I've quite a few very active projects based on nevow/athena, (operating at 100k hits/day) dubbing nevow/athena as 'a large pile of aging JavaScript' sounds like it's going to be thrown overboard and simply does not credit who's credits are due.
Just to stir the pot, with a broad userbase and still having 30% of the users sitting at IE6, nevow/athena is THE single bidirectional framework which deserves this name and spans everything up to the most recent implementations although there are some bug fix reviews outstanding in the for me important part of attaching/removing elements.
In my opinion there is nothing really comparable to nevow/athena out there (may be node.js is slowly evolving in that direction) adoption of RIAs is painfully slow, people are content with 'search proposal' type pages and take it for granted that every highly interactive browser based app is Flash driven.
Writing RIAs was what fed me for last few years so if nevow/athena is going to be flushed count me in as able to host it on trac/svn
Werner
On 12/20/10 3:06 AM, exarkun@twistedmatrix.com wrote:
On 01:15 am, foom@fuhm.net wrote:
On Dec 19, 2010, at 7:25 PM, Glyph Lefkowitz wrote:
Currently the plan is to move the resource model [..]
You're kidding, right? :(
I think "resource model" is a pretty vague way to explain the plan (the next phase of the plan is actually to identify specifically (and _write down_) what should be copied into Twisted Web). Here are some specific things that Twisted Web should adopt from Nevow:
- locateChild
- Deferreds in Resource lookup
- Deferreds in response rendering
- Something like `NotFound`
Probably some other things too. Are these the things about "resource model" that make you sad? Or did the phrase conjure something else?
Jean-Paul
Twisted-web mailing list Twisted-web@twistedmatrix.com http://twistedmatrix.com/cgi-bin/mailman/listinfo/twisted-web
On Mon, Dec 20, 2010 at 8:37 AM, Werner Thie werner@thieprojects.ch wrote:
I've quite a few very active projects based on nevow/athena, (operating at 100k hits/day) dubbing nevow/athena as 'a large pile of aging JavaScript' sounds like it's going to be thrown overboard and simply does not credit who's credits are due.
Well, considering it's one of the authors of it saying that, I'll accept it from whence it comes. And I agree to a large extent, Athena could be better in many regards, and twisted.web2 could probably have helped that.
Athena can make life quite painful if your deployment environment isn't the perfect Divmod way, which is understandable in the context that it was created, and most other ways are indeed wrong. That's why I did a lot of work on Enamel to wrap some of Athena's somewhat laborious bootstrapping requirements.
In my opinion there is nothing really comparable to nevow/athena out there
It's more a case of "It works now for many people, so great, let them continue using it as it is" but as for the /future/ of the Divmod projects, bugs and feature requests, one might as well say "If it can be better, then there is opportunity to make it so".
On Dec 20, 2010, at 1:37 AM, Werner Thie wrote:
I've quite a few very active projects based on nevow/athena, (operating at 100k hits/day) dubbing nevow/athena as 'a large pile of aging JavaScript' sounds like it's going to be thrown overboard and simply does not credit who's credits are due.
Sorry, this was not intended to be a comment on the quality of the implementation. Although the implementation does have some issues, I share your impression that "nevow/athena is THE single bidirectional framework which deserves this name".
I was trying to comment on the challenges presented by maintaining Athena. JavaScript is probably the most popular programming language on earth at this point by some metrics, with new techniques and new libraries popping up every day, and since JavaScript actually have the concept of a "library", each one seems to require some new hack from Athena to work right.
Not to mention more browser versions popping up every day. Browser interactions are complex and the mocking layer used to test Athena is incomplete.
Maintaining Athena's code requires a skill set almost wholly unlike every other area of Twisted, the rest of which is all in Python (with a tiny bit of C) and all relatively easy to test by simply running the unit tests. Not to mention the fact that twisted.web is _already_ strained for lack of maintainers, and I'm sure that the code that we already want to add (the template system) is not going to make that problem any easier.
Writing RIAs was what fed me for last few years so if nevow/athena is going to be flushed count me in as able to host it on trac/svn
It's not going to be "flushed". My comments were an explanation of why Athena will likely remain a separate project from Twisted, not why it is going away.
If you want to do something, hosting trac or SVN is probably not the best thing - we can take care of moving it to a project hosting site. Focus on continuing to maintain the code, removing deprecated things, and simplifying the implementation so that it can be more easily maintained going forward. I personally haven't had time for Nevow in quite a while and I'm not likely to be free to hack on it in the forseeable future.
I would certainly feel differently about Athena+Twisted if there were a large community of active maintainers who were working diligently to gradually remove all the deprecated stuff and strip it down into a simple, coherent core.
If you'd like to host something, a buildbot would be the best. I don't think we have a plan for bringing that back just yet.
Hypothetical future stuff aside: today, right now, Twisted generates some HTML in a few places. It does this in a variety of ugly, potentially buggy ways. There are many simple features I can think of which haven't been added to twisted.web because generating more HTML without adding a pile of new dependencies would be so unpleasant. I'd like it if it could do this using a more robust technique, which is why I'd like to see Nevow's templating system in Twisted: it supports Deferreds, and (modulo compatibility cruft) it's pretty simple.
On Dec 20, 2010, at 2:09 PM, exarkun@twistedmatrix.com wrote:
On 06:35 pm, glyph@twistedmatrix.com wrote:
day, and since JavaScript actually have the concept of a "library", each one
... JavaScript *doesn't* actually ... ;)
Jean-Paul
Apparently this fact is so scarring I can't even bring myself to write it! :)
I admit I never stepped forward doing even review on work on twisted being usually overly busy with maintaining the usual mayhem of HTML, CSS, JScript and Python in my projects. Now with nevow/athena and qooxdoo merged I might have some spare time not having written a single line of HTML or CSS in the last six months and still producing a stream of really advanced and great looking RIA stuff with precision and ease.
The lessons I learned again from merging nevow/qooxdoo
1. the productivity breakthru in writing RIAs comes from shedding HTML and CSS altogether and in concentrating to two full blown programming languages (read Javascript, the good parts - Douglas Crockford)
2. one might even push the above said further by concentrating to one language only by using JScript on the server too, although node.js has a long way to go
3. divmod/nevow does a prime job in bringing an 'as pythonuesque' way as possible to JScript, meaning point 2. is mute, because the server side with twisted is really outstanding and is a beacon among the server programming frameworks
4. it might be nice to have 'whip it up fast' templates and tools but in the end it's the 'finish it faster and maintain it with less work' attributes which let's you make a living. It's always worthwile to invest into something as big and all encompassing as twisted even in a time where the general attitude is 'take this, add hot water and stir it'.
Although it might have been said a hundred times already, but how do I do that intimidating step to become a reviewer in the twisted community?
Werner
On Dec 21, 2010, at 5:23 AM, Werner Thie wrote:
Although it might have been said a hundred times already, but how do I do that intimidating step to become a reviewer in the twisted community?
Mostly, you just ask. We haven't said "no" to anybody yet :).
Also, I think the list of reviewers for Twisted stuff and the list of reviewers for Divmod stuff is different, although I guess that might change once we move the Divmod projects to launchpad. We may ask you to be the author on a few tickets first, to demonstrate that you understand the review process. Of course, it would just be generally nice if you'd fix a couple of tickets anyway :).
Reading http://twistedmatrix.com/trac/wiki/ReviewProcess is a good start.
-glyph
Just to give another voice for Athena - it is the best, I have found and Werner Thie confirmed my choice. I used it in the past successful an would like to use it in future again, so it would be great, if Nevow/Athena keep living and maintained.
Sorry for my English, merry Christmas, Paul
On Sun, Dec 19, 2010 at 8:15 PM, James Y Knight foom@fuhm.net wrote:
On Dec 19, 2010, at 7:25 PM, Glyph Lefkowitz wrote:
Currently the plan is to move the resource model [..]
You're kidding, right? :(
I'm not quite sure what you mean by this, I didn't expect any surprise. I'm just referring to the collection of long-planned stuff that everyone already knows about, like Jean-Paul said. There are already tickets for most things, like accepting Deferreds from Resource.render and Resource.getChild (http://twistedmatrix.com/trac/ticket/3621 and < http://twistedmatrix.com/trac/ticket/3711%3E), and using locateChild rather than getChild (http://twistedmatrix.com/trac/ticket/3763); we should really collect them into one place.
You could also almost as accurately say that we are moving twisted.web2's resource model into twisted.web; it got most things right. Last time I looked at this, I recall that Nevow fixed some mistakes but also introduced some new ones of its own, so ultimately I hope we finally end up with an interface that takes the best from all of these options, and doesn't look exactly like any of them.
On Mon, Dec 20, 2010 at 2:25 AM, Glyph Lefkowitz glyph@twistedmatrix.com wrote:
We're not planning to move Athena into Twisted, as it's a large pile of aging JavaScript, and would involve a significant amount of browser-based testing in order to get right, which Twisted doesn't currently need.
Athena would need to be stripped down to a much smaller core in order to be appropriate for this.
Beg your pardon ladies and gentlemen, may I my 2¢ here?
I would be totally happy if the result, in the end, looked almost like Perspective Broker (given that it is fairly decently documented right now, no need to reinvent wheels) on the Python side and would be easy to hand-write in pure JavaScript on the front-end side (so that you ultimately don't necessary need generated magic code your average front-end teammate proficient in JS but vaguely familiar with Twisted and Python in general would have to dig into and work around).
I think that documentation for the server side would be rather a simple comparison of what things doable with PB cannot be done with Athena. The only piece to have a deeper write-up would be the JavaScript side.
If that was the case, I would seriously look at bridging that with jQuery.
On 12/20/10 3:18 PM, Yaroslav Fedevych wrote:
On Mon, Dec 20, 2010 at 2:25 AM, Glyph Lefkowitz glyph@twistedmatrix.com wrote:
We're not planning to move Athena into Twisted, as it's a large pile of aging JavaScript, and would involve a significant amount of browser-based testing in order to get right, which Twisted doesn't currently need.
Athena would need to be stripped down to a much smaller core in order to be appropriate for this.
Beg your pardon ladies and gentlemen, may I my 2¢ here?
I would be totally happy if the result, in the end, looked almost like Perspective Broker (given that it is fairly decently documented right now, no need to reinvent wheels) on the Python side and would be easy to hand-write in pure JavaScript on the front-end side (so that you ultimately don't necessary need generated magic code your average front-end teammate proficient in JS but vaguely familiar with Twisted and Python in general would have to dig into and work around).
I think that documentation for the server side would be rather a simple comparison of what things doable with PB cannot be done with Athena. The only piece to have a deeper write-up would be the JavaScript side.
If that was the case, I would seriously look at bridging that with jQuery.
Might be a matter of preference, don't you think?
I'm running on a merge of nevow/athena with qooxdoo bringing RIA development really to live.
Just to throw some figures around, rewriting my card game platform with that approach I was able to shed about 70% of the code on the client side, not a single line of HTML or CSS - plain, clean JScript instead, with browser compatibility across every thing existing starting from the times IE6 ruled. Server side code shrinkage with isolating game logic with pb is around 60%, the whole rewriting took about 1/10th of the original implementation time, meaning this was the biggest gain I've ever seen in productivity since I started dabbling with code in 1972.
All in all, I'm probably already stuck with nevow/athena, but would commit to it any time again. So let's make nevow/athena better.
Another 2cts, Werner
On Dec 20, 2010, at 9:55 AM, Werner Thie wrote:
with browser compatibility across every thing existing starting from the times IE6 ruled. Server side code shrinkage with isolating game logic with pb is around 60%, the whole rewriting took about 1/10th of the original implementation time, meaning this was the biggest gain I've ever seen in productivity since I started dabbling with code in 1972.
All in all, I'm probably already stuck with nevow/athena, but would commit to it any time again. So let's make nevow/athena better.
Another 2cts, Werner
+1 on all the above. Plus every time I realize my AJAX app runs on IE6 automagically I get a little joy explosion inside ;-)
-phil
On Mon, Dec 20, 2010 at 4:55 PM, Werner Thie werner@thieprojects.ch wrote:
Might be a matter of preference, don't you think?
Or, a matter of coexistence, while we are at it.
On 05:52 pm, phil@bubblehouse.org wrote:
I'd like to add that I'm a bit more concerned about divmod.org accessibility, especially now that I've got a project that depends on Nevow.
I put a tarball of Nevow-0.10.0+r18026.tar.gz on my download site for now, but I'm glad to do anything I can to help get the repo back up.
If there's no interest in hosting the repo, lets move it to launchpad or github or something. If those concerned have no time, I'm glad to steward the Nevow source for awhile.
We're working on getting the source onto Launchpad.
Jean-Paul
participants (8)
-
Colin Alston -
exarkun＠twistedmatrix.com -
Glyph Lefkowitz -
James Y Knight -
Paul Reznicek -
Phil Christensen -
Werner Thie -
Yaroslav Fedevych