--
You received this message because you are subscribed to the Google Groups "Joomla! Framework Development" group.
To post to this group, send an email to joomla-dev...@googlegroups.com.
To unsubscribe from this group, send email to joomla-dev-frame...@googlegroups.com.
For more options, visit this group at http://groups.google.com/group/joomla-dev-framework?hl=en-GB.
Amen to that.Although I think mercurial would be easier to pick up by the majority (specially windows users), bitbucket is not bad at all in comparison to github, and I'd say more straight forward to start using.
As I understand it, there is no enthusiasm among the Joomla! Dev Coordinators for moving away from SVN.
The only issue I've run into with the SVN-Git bridge is that the patches produced by Git are not compatible with SVN. If you're committing back to your own branch on JoomlaCode, this isn't a problem. However, if you don't have commit access, you'll have a hard time sharing your patches in the tracker.
I haven't looked into Mercurial as seriously as I'd like to. Definitely wish Git was more Windows-friendly (although that doesn't affect me directly).
-Joe
I use both git and mercurial, on the basic level they are practically the same (git has more toys to play around though), but I would be concerned if the Joomla dev community had more windows than linux/mac devs, does anyone have any idea on what the proportions are?.
@JosephIf I'm not mistaken, Joomlacode already supports git, and because of git's distributed nature, Joomlacode could play along github with no problem.
@allWe can debate all day wether the project should leave svn in favor of a DVCS, however, our opinion alone won't achieve anything and some may even say "things are good how they are, why move?", even with facts on the table. I'd recommend we do the following:- Make a pros/cons list on the switch to a DVCS- Make a DVCS comparison according to the needs of the project and community- Make an action planPerhaps we could take as an example how the Drupal project handled this very situation: http://drupal.org/community-initiatives/gitEven if we get a "no" (because I'd guess the final say is of the project's committers), I think a switch to a DVCS on the long term would be inevitable and at least we could lay something out for the future, or we could even come to a plan where svn can coexist with no problem (for example, an automated script to update an official git repository from the official svn repository).
And yes, we could "bridge" Joomlacode and github as Gary is suggesting. So long as people who want to test patches are willing to check out using Git, we won't have the problem with patches. However, if you try to generate a patch and post it on the tracker, you'll end up with a format that's incompatible with SVN.
My concern isn't so much programmers as much as Bug Squad testers. But then again, I don't have any hard numbers on this. Perhaps we should do a Bug Squad survey?
I see a Git plugin for GForge, but it doesn't look like it has full support: http://gforge.com/gf/project/scmgit/
And yes, we could "bridge" Joomlacode and github as Gary is suggesting. So long as people who want to test patches are willing to check out using Git, we won't have the problem with patches. However, if you try to generate a patch and post it on the tracker, you'll end up with a format that's incompatible with SVN.
That sounds like a reasonable way forward. Should we continue this on a people.joomla.org group perhaps?
That sounds like a reasonable way forward. Should we continue this on a people.joomla.org group perhaps?Not sure, I think the Google Groups have more visibility but I could be mistaken. I could create the page at the wiki and we could get comments on both ends perhaps?.
--
Interesting stuff. My first impression before going into the pros and
cons on various systems is actually establishing what we are comparing
to. First I'd be establishing the requirements for what we actually
need. Then you can do up a decision matrix and see what comes out in
the wash. For example, the advanced tools in Git will mean nothing if
it doesn't support a platform we require to be available to the JBS.
Also the needs of the core repo and that of options for creating
branches are two separate issues (the latter is obviously far more
flexible).
Those are just my initial thoughts anyway.
Regards,
Andrew Eddie
http://www.theartofjoomla.com - the art of becoming a Joomla developer
For example, the advanced tools in Git will mean nothing if
it doesn't support a platform we require to be available to the JBS.
For what it's worth, I'm *very* keen to look at Mercurial.
Off the top of my head, things that we "need" to look at for the
source tree are:
* Availability of Window GUI (eg, similar to Tortoise)
* Command line binaries for OSX
* Command line binaries for the usual linux suspects
* Plugins for Eclipse mandatory; plugins for Netbeans and Zend Studio
highly desirable.
* Patching procedures no harder (preferably easier) than SVN
* Resources required are no harder to set up than we have now for SVN
When you go past the source tree and into the collaborative
development area of the project, I'd have no qualms about looking at
supporting 3 branching repos: SVN, Mercurial and Git for those that
want to work in those spaces providing we have people with expertise
to support them and help contributors, and we have good
bridges/procedures (I'm not sure what the right terminology is here)
to get the goods back to the source tree. We should certainly let
people innovate on the system that they are most comfortable with.
Regards,
Andrew Eddie
http://www.theartofjoomla.com - the art of becoming a Joomla developer
Hi David
For what it's worth, I'm *very* keen to look at Mercurial.
Off the top of my head, things that we "need" to look at for the
source tree are:
* Availability of Window GUI (eg, similar to Tortoise)
* Command line binaries for OSX
* Command line binaries for the usual linux suspects
* Plugins for Eclipse mandatory; plugins for Netbeans and Zend Studio
highly desirable.
* Patching procedures no harder (preferably easier) than SVN
* Resources required are no harder to set up than we have now for SVN
When you go past the source tree and into the collaborative
development area of the project, I'd have no qualms about looking at
supporting 3 branching repos: SVN, Mercurial and Git for those that
want to work in those spaces providing we have people with expertise
to support them and help contributors, and we have good
bridges/procedures (I'm not sure what the right terminology is here)
to get the goods back to the source tree. We should certainly let
people innovate on the system that they are most comfortable with.
Thanks for that - that's an awesome link resource. My present
understanding of things is both Git and Mercurial would be better than
SVN, but for the master source tree, Mercurial is probably as you say
a better fit (or a lower common denominator if you like).
Regards,
Andrew Eddie
http://www.theartofjoomla.com - the art of becoming a Joomla developer
I think regarding Windows users there will always be "enough" to have
to support that at a central project level.
Don't suppose you can make it to San Jose in October?
Regards,
Andrew Eddie
http://www.theartofjoomla.com - the art of becoming a Joomla developer
> That is correct, either Git or Mercurial are better than SVN, I guess that
> any distributed SCM should be specially better than any centralized SCM when
> it comes to big projects (any important open source project).
FullACK.
> As for the
> master repository I think mercurial would be the best pick, but I think it
> would really depend on how many Windows users we have in the JBS as Joseph
> pointed out earlier.
Since git is supported by Eclipse and NetBeans, which both are platform
independent, will that not be sufficient Windows support?
Wondering
Niels
--
| http://www.kolleg.de · Das Portal der Kollegs in Deutschland |
| http://www.bsds.de · BSDS Braczek Software- und DatenSysteme |
| Webdesign · Webhosting · e-Commerce · Joomla! Content Management |
------------------------------------------------------------------
Since git is supported by Eclipse and NetBeans, which both are platformindependent, will that not be sufficient Windows support?
Any plan to move to an alternative (D)SCM would have to include a
migration strategy that included these items as well.
Cheers,
Sam Moffatt
http://pasamio.id.au
The current release of JoomlaCode has GIT support however if we were
to consider shifting that way we'd need to convert all of the
releases, branches and history in a uniform GIT repository so that we
don't lose anything. The JC SVN repository also has more than just the
development tree, there are other trees there as well.
>> Since git is supported by Eclipse and NetBeans, which both are platform
>> independent, will that not be sufficient Windows support?
> The GUI client support is there, there is a TortoiseGit project and
> everything. It's the server stuff that craps out on windows and makes
> it infeasible.
Ah, I see - thank you for your explanation!
Regards,
Niels
-Joe
On Aug 17, 2010, at 5:45 PM, Gary Mort wrote:
> BTW, my own analysis has led me to use the following:
> I use git as my primary SERVER DCVS because it has better merge tracking capabilities and is more popular.
>
> I use mercurial as my LOCAL dcvs because it has better windows support. Mercurial will easily pull/push code from a Git server, so there is no need for me to limit myself to the DCVS with the better windows client, I can use both where they are appropriate.
>
>
>
>
>
> Ok, as a non-Windows Git user, this clears up some questions in my
> mind. So if you can use Mercurial as your client on Windows to talk
> to a Git server elsewhere, I'd strongly suggest we move to Git for
> the central repository rather than Mercurial.
+1
Regards,
Niels
> I think the optimal setup would be the main repository to be in joomlacode,
> pulling/being-pushed-to from github and bitbucket.
Sounds very reasonable to me, provided, that the synchronization will be
transparent for the users of the respective repository.
+1
Regards,
--
--
> Unfortunately, it seems like the patches
> generated by Git are incompatible with the typical UNIX patch
> command.
Is there any reason, why you don't create the patch from the IDE?
NetBeans does support that...
Regards,
Niels
Have you tried generating a patch from a diff of a Git repository in NetBeans, then applying that patch elsewhere? I'd assume that NetBeans and Eclipse would still have to implement some sort of logic to convert the diff from Git into one that's compatible with what the rest of the UNIX world uses.
-Joe
> Have you tried generating a patch from a diff of a Git repository in NetBeans, then applying that patch elsewhere? I'd assume that NetBeans and Eclipse would still have to implement some sort of logic to convert the diff from Git into one that's compatible with what the rest of the UNIX world uses.
AFAIK NetBeans (and Eclipse) (can be configured to) use the external
diff command. Can't check this currently.
Regards,
Niels
-Joe
> and one which purports to be a solution:
> http://www.mail-archive.com/d...@trafficserver.apache.org/msg00864.html
This is the original source:
http://mojodna.net/2009/02/24/my-work-git-workflow.html
Regards,
Niels
To unsubscribe from this group, send email to joomla-dev-frame...@googlegroups.com.
To unsubscribe from this group, send email to joomla-dev-frame...@googlegroups.com.
Hi David. That's a great coincidence. Recently, Louis suggested and the PLT agreed to use bitbucket & mercurial for the new platform repository. So the platform project will use mercurial/bitbucket instead of svn/joomlacode from the start. This is expected to be set up and working on or before April 15.
The plan is to use mercurial for a period of time with the platform project. If it works out well (which we hope and expect), then we would consider changing the CMS project over to the same repository.
I'm not exactly sure the time frame for changing Joomla to mercurial. Perhaps run the platform project for 6 months or so and then evaluate? Not sure.
In the meantime, I'm also not sure the best way to proceed on the idea of a bitbucket/mercurial "mirror". If there is a way to set this up soon without a huge amount of work, I think it would be most interesting to try. I'm not sure about the mechanics. The Joomlacode SVN would continue to be the "real" repo for the cms in the meantime, but having the bitbucket as an option would be valuable to help with the long-term transition.
Anyway, that's where things stand. Louis, Sam, Ian: Any other thoughts about this?
Thanks. Mark
On Mon, Feb 21, 2011 at 6:20 AM, dukeofgaming <dukeof...@gmail.com> wrote:
Hi Mark,
Just to ask if there is a repository set already, if not, just to
suggest using bitbucket (https://bitbucket.org/) for it. Also, I'm
adding this development to the wiki page:
http://docs.joomla.org/Dvcs#Action_Plan
Regards,
David
We have been working for a month (maybe some wheel spinning) on a Mercurial extension that will help with the database, we are trying to use Fog Creek's Kiln for multiple-person php development on locals and to track changes, discrepencies, etc. especially, when it comes to doing work on a live website (from local) and not losing live changes. (and for documentation, time tracking, etc.) I'm not sure if what we have been working on is helpful or even relevant to the bigger picture of Joomla core development... I suppose that we could help with posting some documentation, where would be a good place for my team to share some of our ideas and learn from other people's progress on using Mercurial for Joomla?
Thanks for your understanding.
Regards,
Andrew Eddie
http://learn.theartofjoomla.com - training videos for Joomla 1.6 developers
Well, technically the effort happens anywhere and any improvements to
the framework are still under the auspice of the ideas forum and
Joomla feature tracker. I think it's been previously mentioned that a
few of us are experimenting with distributed version control
technologies. There have been quite a few discussions about which
technology to use but the consensus seems to be to start with
Mercurial because it is the most convenient for Windows users. To
that end, trials are being set up on Bitbucket. I can only speak for
myself, but so far it looks promising but I have not tooled up on it
sufficiently to give any authoritative guidance beyond "it looks like
this way might have legs". I'm impressed enough to start moving my
client work to Bitbucket to help me learn the system better.
> - Does the CLA cover the platform project?
I can only speak for myself but I signed the CLA to cover my
contributions to the Joomla project, not what technology my
contributions are committed to.
> - Will the framework be distributed at joomla.org?
I can't think of a reason it would not be, at least initially.
Though, speaking for myself, developer.joomla.org may be an
appropriate long term home for it. Don't really know, and to be honest
I don't really care, but I'm sure wherever the download point is, it
will make sense.
> - Is there documentation that defines "what the framework is?" In other
> words, how it's being separated from the CMS? (Is it just the joomla!
> libraries folder? or more?) (If there is no documentation, that's fine, but
> if there is it would be excellent reading).
The framework is everything under /libraries/ and the goal of any
framework effort would be to progressively decouple as much as
possible from the "CMS". Obviously any first steps need to be
"gently, gently". I could also see the CMS "adding" additional
libraries for it's own purposes that are coupled to supporting CMS
"stuff" but not really appropriate in the core framework.
> - How can community participate? (Can we share bug reports and patches? For
> example, JForm improvements. If so, where and how? Or, is involvement
> planned for a later point in time after the framework is separated?)
As I said before, contributing to Joomla is not dependant on the
technology we use for source management. I believe I and others have
explained in depth the process of how contribution works via the ideas
forum, these mailing lists and the Joomla feature tracker. Mercurial
is just a "tool" that makes it easier for people to manage their
efforts, and would form the point of truth for at least the framework
as I understand it. There is nothing precluding anyone from forking
(that's DSCM speak) their own framework or CMS repo using Git, Bzr or
even SVN on Joomlacode. All that's happening is a different SCM is
being looked at for the framework project (and if that works, maybe
the CMS will shift as well in the long term) - it doesn't change the
systems already set up for contributing (just the final destination
for committing accepted work). At least, that's the way I see it.
> - What license will be used for the framework? (In the past there was a lot
> of talk about possibility using the LGPL but I don't believe a final answer
> has been shared.)
I don't believe a final answer exists to that question other than the
CLA would allow the project to choose the LGPL in the future if it so
desires, which is ok with me otherwise I wouldn't have signed it.
To sum up, as I see it, it's early days yet but whatever systems are
already in place will logically extend to the platform project, at
least in the interim. If change is needed in the future, that bridge
can be crossed when needed.