I completely agree with everything you said... This project has so much promise and what has been done so far is excellent and Kudos to Luke et al. for getting it this far but to move forward and to make sure other users aren't put off, we really need to move in a more practical way.
I myself spent about 10 times longer than I would normally on a project like this only because I REALLY like Python and was willing to look past the warts and difficulties in getting set up. Alot of people (me included) will go to a project website look briefly at the blurb download and want to be working straight away and if not, just give up an find one that is that easy. I know you shouldn't EXPECT free software to be perfect but alas, life is too short and there are loads of projects that are easy to get setup, with accurate information etc. On Thu, Dec 8, 2011 at 5:48 PM, Pascal Chambon <[email protected]> wrote: > @ Lex & Apexi & Luke > In PRACTICE that sourceforge repository is the main one, but in THEORY it > was marked as the legacy one, as noted in the right bottom corner of > http://pyjs.org/. Sure master branches of the two are in sync, but other > branches aren't at all. And I've found cute stuffs on code.google.com, > that hadnt been transferred to any of these two repos. Time to settle > things, isn't it ? > > > > > Really, for newcomers, pyjamas is everything but friendly, sources and > > > tools are scatered/duplicated/triplicated all over the web, and there > is > > > no central, up-to-date manual on how to develop and test on it... > > > yes there is - it's DEVELOPER.RULES and also, there is a general > > expectation that if there's something you don't understand, it's best > > to ask on the mailing list. that's how free software projects work. > > Even reaching the mailing list is an achievement here, and I guess most > developers don't > know exactly what are the bits of pyjs you can find here and there... I > guess a real centralization, and a little, > very practical manual would save a lot of time. > > > > > Fixing this wouldn't be an incredible job really, it just requires a > > > "green light" from core devs (and some admin credentials for me if > > > nobody else wants to do it) > > > works for me. hmmm... i've got an ssh public key for you already, > haven't i? > > I thought this key was only for the git repo, does it give me admin rights > on the whole site ? I'll try at home. > > > > > So here are my propositions: > > > - make deprecated repository(ies) readonly, > > > there aren't any. the one on pyjs.org is "sync'd" - i keep meaning > > to shut down the one on sf.net, just keep not getting round to it. > > pyjs.org repo only confuses stuffs, it lacks all branches and a proper > browser, please - let's let sourceforge handle that scaffolding since > they're completly expert at it. I vote for the shutdown of pyjs.org repo. > > Also, it's not explicit enough that http://code.google.com/p/pyjamas/ is > COMPLETELY deprecated, how about shutting it / marking it on the front > page, so that people dont think that it's still used for any task ? > > > > - cleanup and update the pyjs site, which references deprecated depots, > > > mailing lists, and has broken links for important pages like "supported > > > cpython features". Why not speak about 0.8alpha in it ? > > > because i work on pyjamas when *i* need it. and, given that i have > > not had any major paid-for contract work (web sites etc.) that require > > pyjamas, i am operating in a much more "passive" role. > > > if on the other hand there was a rush of people requiring pyjamas > > contract work, i would immediately have the time, money and incentive > > - personally - to do things like update the web site. > > > however, given that there is not, i am keeping my role to an absolute > minimum. > > I perfectly understand that, and that's why I dont ask you to do anything > (I'm already quite impressed by all what has been done on that project), I > just ask for an agreement on what should be done. If I have a "green > light", then I should be able to handle most of this cleanup (always easier > to remove stuffs than to add them, isn't it ?). > > > > Pyjs is for > > > dynamic applications, not webpages for which pure html is 100% fit. > > > it's not a panacea. you have to follow the "AJAX search" rules in > > the server back-end. there was some discussion about how google had > > added ! and > and some half-baked scheme into their search engine in > > order to index AJAXified sites. a pyjamas site is no different. you > > just have to follow the rules. > > Replacing ajax tabs by mere html pages would really make things easier and > more robust in that case, isn't that so ? I'd personally reserve pyjs pages > for truely dynamic appz, that don't even have the notion of "static urls" - > eg. administration backends, utilities etc. > > > > - use more robust tools: we have a mailing list that doesn't handle > CCs, > > > if you'd like to do an update to lamson, that will get taken care of. > > however you have to watch out because i had to fix some bugs in > > lamson. > > Well I don't know Lamson, but does having a python-based mailing list > truely help the project ? Mainstream lists like google's require absolutely > no work from us, so that we can concentrate on main goals, not fixing > secondary tools. > > > > a source browser which is cute but doesn't have half the feature of > > > mainstream source browsers... really in some case "eating your own dog > > > food" is a hindrance, > > > sorry, but tough. read richard stallman's essays on the principles > > behind free software. if it's "monetarily zero cost" it's *NOT* > > Software Libre. i'm a Software Libre Developer, not a banged up > > wannabe pretender looking to screw people over in the name of > > maximising profits. > > So you mean it's actually a political issue, not a technical one. But I > don't understand, aren't there already lots of GPL mailing lists, source > browsers, etc. ? And even if mainstream tools I think about don't pass your > "software libre" requirements, does it mean we must content ourselves of > handmade, constantly deprecating tools ? > > Personally I think that's where "practicality beats purity". When a hammer > fits my need, I don't demand the hammer factory and the plans that were > used to create it. I don't want to hack mailing list systems, nor > versioning systems, nor source browsers, I just want to USE them as BLACK > BOXES, in order to be able to easily hack into pyjamas, and share infos > about it. > > So imo the only criterium is the featurefulness and the robustness of > these tools, not what license they come with. At the risk of being damned > forever, I even assert that I'd gladly install a costless "all in one" pack > from microsoft on pyjs.org, if it fixed instantly all the flaws I'm > ranting about :p > > > > the only reason i haven't shut down the use of code.google.com and > > the sf.net sites is lack of time to create alternatives that are > > pyjs-based. > > tell me where i can get the source code for such key-in-hand > > solutions, under Software Libre Licenses, written in python and better > > yet designed such that a pyjamas front-end can easily be slotted on > > top of them and i will happily replace them. > > So when you work on a framework, you want the entire world to be rewritten > using that framework ? I must not have understood properly ^^. > There are tons of FOSS projects thriving on sourceforge, why on earth > should pyjs do stuffs differently (a fortiori when we don't have millions > of hours to invest into maintaining tools ?) ? > Let's be pragmatic, I enjoy open-source as much as you, but I certainly > won't die for it, nor constrain my projects with Stallman's dogmas. And > it's not because I love python that I won't joyfully integrate javascript > plotting libraries, ruby project managers, and even PHP mailing lists, into > my workflow, if it avoids me to waste 10000 hours of dev time. > > So how about deploying a very central, common, robust infrastructure for > the pyjs project, with tools that other projects maintain for us, so that > every newcomer has a single point of information ? Isn't the benefit worth > the "purity loss" ? Or shouldn't we do a vote about it, as any "Libre" > community should do ? > > Regards, > Pkl > >

