On 11/21/2012 09:15 AM, Vincent Massol wrote: > Hi Guillaume, > > I'll stop the discussion here because I don't find it very interesting. > > Let's just conclude that in meritocratic open source project the people who > decide on the project are the doers… So if you want something, just do it and > show the way... > > If you're not a doer, you can always whine but it won't do much…
Well said, Vincent. > Thanks > -Vincent > > On Nov 21, 2012, at 3:08 PM, Guillaume Lerouge <[email protected]> wrote: > >> Hi again, >> >> On Wed, Nov 21, 2012 at 2:58 PM, Vincent Massol <[email protected]> wrote: >> >>> >>> On Nov 21, 2012, at 2:48 PM, Vincent Massol <[email protected]> wrote: >>> >>>> On Nov 21, 2012, at 2:33 PM, Guillaume Lerouge <[email protected]> >>> wrote: >>>> >>>>> Hi Vincent, >>>>> >>>>> if I may, this looks like a common fallacy: developers wanting to build >>>>> tools for developers. Of course building a "XWiki for Software >>> Development" >>>>> flavor will sound sexy to you, since you're a developer yourself as >>> well as >>>>> a XWiki committer. You would be your own target audience. In other >>> words, >>>>> you want to build something for yourself. >>>> >>>> I think that not being a developer you completely miss the point :) >>>> >>>>> However, please note that the market for such tools is already very, >>> very >>>>> crowded. >>>> >>>> Market? Who's talking about marketing/research studies, etc here? :) >>>> >>>>> There's Trac / Bloodhound, there's the whole Atlassian suite, >>>>> there is what Github is building as well as countless other solutions. >>>>> >>>>> One of XWiki's great strengths and differentiators is in its ability to >>> let >>>>> people manage structured and unstructured content easily. I think we >>> should >>>>> keep focusing our work on this instead of trying to enter a crowded >>> space >>>>> with little perceivable benefits >>>> >>>> Who's "we"? >>>> >>>>> . In your mind, is this the very best thing >>>>> we could possibly work on in order to ensure XWiki's long-term success >>> and >>>>> sustainability? >>>> >>>> Definitely. >>> >>> Just to explain: what we need for xwiki's long term success are >>> contributions and who does contributions? Developers. Appealing to them is >>> needed. Right now I've never succeeded in getting any developer interested >>> in XWiki by presenting it to them. I believe that giving them a tool they >>> want to use for their job would make XWiki more appealing to them. >>> I agree with Vincent. This has also been my vision for a very long time. And yes, this is *definitely* the *best* thing that we can do to get XWiki popular/famous, and as a consequence, to get more customers for all the companies that want to base their business on the XWiki open source products. Why? Well, XWiki SAS has been pursuing the same business model for many years, and while it did grow, I've seen many more open source startups grow exponentially into multi-million businesses in far shorter times, and they did that by being popular in the developer community. >> Following this type of logic, if you were working on Microsoft Office you'd >> start adding code-highlighting features in order to make it more appealing >> to developers. Sounds like a great idea, doesn't it? That's a lame analogy. Although some say that "XWiki is the Excel of the Web", what does the XWiki Platform have in common with an Office suite? I've always presented XWiki as a "web application development platform", and not as a "wiki engine". And a development platform needs development features. Feel free to view XWiki any way you want, even as a MS Office alternative that doesn't need code highlighting, but keep in mind that you're just one member of the community, and you don't exclusively control the vision of the project and the direction in which other community members want to move. To counter your example even more, MS did go even deeper than code highlighting in MS Word: they developed an open source (!) Office extension for AsmL, which allowed researchers to write formal specifications inside Word documents, and those specifications could be validated and compiled directly from within Word. Why would they do that? The official reason for embedding such a feature in Word was that researchers will write such formal models in their publications anyway, so why copy/paste them from an external tool when they could write and validate the models straight in their paper, and reviewers could just validate the models straight from Word? This wasn't a move toward attracting a much larger user base than what Word already had, it was a move with network effects in mind targeting a very very specific userbase, but with large ripple effects. It was a move toward keeping researchers using MSOffice, and transitively Windows, instead of other open source tools that didn't require any MS products, and researchers and other academic personnel in turn influence a lot of people. >> What matters for the long-term success of any piece of software is to have >> users who love using it. Those users do not necessarily have to be >> developers. Developers tend target platforms where they know they will be >> able to reach numerous users. So I'd go about this the other way: build a >> platform that people love using, and developers will come. Let me remind you that most of the users that you're targeting are simple users of a customized enterprise solution built by XWiki SAS employees, hiding all the development features behind easy to use interfaces. Those kinds of users will rarely be developers. While I agree in general that building a great solution for users will bring developers, that won't happen when you dumb down the solution so that no user will feel the need to tinker with the code, when you hide the fact that there's even the option of tinkering with the code, and when you explicitly oppose the idea of becoming more appealing to developers. How many developers did we get from our enterprise users? I don't remember any. Almost all the community members that actively participate in the community are advanced users, sysadmins or developers that use XWiki as developers, not as end users. XWiki has great features that excited all the other open source developers that I've had the chance to talk with, but nobody knows what XWiki is because so far we've explicitly avoided getting into that "market". Nobody is working on performance because none of our users are big enough to need better performance. Put XWiki behind several high profile communities, and we'll get our performance improved in no time. >> >>> In any case, I'm not proposing some new roadmap or the like. I'm >>> personally going to work on this. Actually I've started working on this as >>> soon as I joined this project several years ago and I'll continue since we >>> need this for ourselves anyway for xwiki.org and that's enough to justify >>> it since we're already spending the time on it. Now with this email I'm >>> trying to get other people interested in this topic (the more the merrier) >>> and to explain what I'm driving at. >>> >> >> Which is exactly why I'm answering. I don't think having additional >> committers start working on this topic (instead of focusing on topics that >> have a direct impact on a large majority of XWiki users, such as search for >> instance) is a good idea. That is your opinion. I think that getting XWiki known in the open source world is the best way to get new users as well. You think that clients can only come from marketing campaigns, while I think that having a product widely adopted by developers is a much better alternative. Did mysql become a billion dollar company by trying to keep open source developers away, and targeting only enterprise users that could afford to pay for it? I see in XWiki the potential to be the next Ruby on Rails, and RoR is a successful web application development framework without enterprise features, yet it still manages to have a lot of monetary value around it. Another aspect is that many times members of other open source communities have a day job in another company, and employees that are good enough to be in an open source community usually have a strong enough voice in their company to be able to suggest solutions such as XWiki. If for the next project they have to do they suggest XWiki instead of Drupal, or if they suggest XWiki Enterprise as a better alternative to their current intranet solution, it's a win for us. At least some of those companies will opt for proper support from one of the companies developing the XWiki software. I agree that not spreading our development resources too thin is a good idea, since we don't really have many developers, but I don't agree that the only "market" for XWiki is the one you see. While the "platform for developers" market is indeed crowded, this is also a highly dynamic market where winners change often based on their merits, and I believe that XWiki has a lot of merits (especially if we fix the performance). Now, to answer Vincent's original email, I do agree with this proposal, and frankly it's one that I've been trying to pursue myself since the beginning, although I've been constantly pushed into other directions. >From now on, I'll work toward making XWiki better for developers. This is also what I'm thinking of proposing as a talk for FOSDEM. Some other tools that we could include: * Poll application for decision making * Presentation application, plus * Event organizing and scheduling (we wrote something for the Community Building and Marketing devroom that I'm co-organizing at FOSDEM) * The l10n application >> Guillaume >> >> >>> FTR here's what we have so far that's finished (and used on xwiki.org): >>> * FAQ application >>> * JIRA macro >>> * IRC Bot (although we still need to fix a few things as Caleb mentioned >>> but they are small) >>> >>> In progress: >>> * MailArchive Application from Jeremie >>> * Release app (see dev.xwiki.org/xwiki/bin/view/ReleasePlans/WebHome). >>> We'll also need to integrate in it the creation of Release Notes as a >>> feature. >>> * Some Git/Github extension: >>> http://extensions.xwiki.org/xwiki/bin/view/Extension/GitHub+Application. >>> I'd like to use it on the Hall Of Fame page on xwiki.org >>> * and probably some more…. >>> >>> Thanks >>> -Vincent >>> >>>> Thanks >>>> -Vincent >>>> >>>>> My 2 cents, >>>>> >>>>> Guillaume >>>>> >>>>> On Wed, Nov 21, 2012 at 1:20 PM, Vincent Massol <[email protected]> >>> wrote: >>>>> >>>>>> >>>>>> On Nov 21, 2012, at 12:40 PM, Jeremie BOUSQUET < >>> [email protected]> >>>>>> wrote: >>>>>> >>>>>>> Hi Vincent, >>>>>>> >>>>>>> >>>>>>> 2012/11/18 Vincent Massol <[email protected]> >>>>>>> >>>>>>>> Hi devs, >>>>>>>> >>>>>>> >>>>>>> >>>>>>>> *** Latest emails (taken from mailman or other mailing list software, >>>>>>>> possibly by subscribing the project to a mailing list so that it gets >>>>>> the >>>>>>>> emails) >>>>>>>> >>>>>>> >>>>>>> >>>>>>>> ** A forum application, for example the Mail Archive Application >>> done by >>>>>>>> Jeremie which would need to be improved to add ability to post from >>> it >>>>>>>> >>>>>>> >>>>>>> Couldn't / shouldn't it be the same thing ? >>>>>>> I know the Mail Archive App is not finished at all, but one feature is >>>>>>> possibility to generate code to include in pages in order to display >>>>>>> filtered lists of emails or topics loaded by the app (filtering by >>>>>>> mailing-list, with ordering, max nb, etc…). >>>>>> >>>>>> Yes, it's the same thing I agree. >>>>>> >>>>>>> If I may add some comment, it's a very nice idea. To me the biggest >>> trap >>>>>> is >>>>>>> integration with external sources. If it's not easily pluggable / >>>>>>> configurable and choice is too restricted, it will attract only a >>> little >>>>>>> subset of developers. In my office for example, I would use it if I >>> could >>>>>>> link to Rhodecode or Mercurial (instead of github) and Redmine >>> (instead >>>>>> of >>>>>>> jira). >>>>>> >>>>>> Yep, we would need contributions for other issue trackers but once we >>>>>> start having something it may attract devs to develop other >>> integrations. >>>>>> >>>>>> Thanks >>>>>> -Vincent -- Sergiu Dumitriu http://purl.org/net/sergiu _______________________________________________ devs mailing list [email protected] http://lists.xwiki.org/mailman/listinfo/devs
