This page is no longer maintained — Please continue to the home page at www.scala-lang.org

GitHub move?

32 replies
gkossakowski
Joined: 2010-03-11,
User offline. Last seen 33 weeks 5 days ago.
Hi,
We (Scala+GWT folks) moved to GitHub recently from Gerrit instance maintaned by us (Rob Heittman to be precise). There are number of things which we like about GitHub but there's one very disturbing: code review process feeling broken.
Let me illustrate what's broken about it. See following pull request:
https://github.com/gkossakowski/testcodereview/pull/1
I did a code review and left some inline comments. We see second submission, we open a commit with a file and we see:
https://github.com/gkossakowski/testcodereview/commit/8dff2a1f7f9187aeee36bf556d96c9368a15d35e
There's absolutely nothing indicating that there were comments left on this file so it's really hard to see a context for this change. It means I have to rescan the whole history of pull request and make notes what has been fixed and what not.
Also, issues raised during code review were addressed by introducing new commits and thus polluting the history. It's not a coincidence. If you rebase your changes and push them again to pull request you'll see the history of pull request being completely broken:
https://github.com/gkossakowski/testcodereview/pull/2
Now, those are only basic problems with code review at GitHub. Other problems include complete lack of tracking dependencies between pull requests.
Now, you might be wondering why I'm writing all of this here, on scala-debate. There are two reasons:
  1. I guess there are many people on this list using github a lot so I hope you can point me in right directions how to solve this problem.
  2. If there are no easy solutions to problems I mentioned then I'll be in position to seriously question GitHub move of Scala. I don't question Git move to Scala, but GitHub seems to add no value and may establish false expectations. Namely, people might think it will be easier to contribute to Scala. I cannot see this happen if there are no good code review tools in place.
To sum up: GitHub seems to be very limited when it comes to handling Google-style code review process. If you dig more you'll find even more issues like that. Given how Scala team busy is I'm afraid that GitHub move won't meet high expectations of the Scala community.
PS. These are solely my views on the issue. I didn't discuss this with either Typesafe or Scala team.
-- 
Grzegorz Kossakowski

Simon Ochsenreither
Joined: 2011-07-17,
User offline. Last seen 42 years 45 weeks ago.
Aw: GitHub move?
Yes, I have hit some of the same problems. Not sure if there are improvements planned from the GitHub people.

I just want to mention that I think that GitHub adds huge benefits, even without that sort of codereview. I personally handled this problem by closing the pull request, fixing the things mentioned in the comments locally, combining it into a single commit with the older changes and doing a new pull request.
Simon Ochsenreither
Joined: 2011-07-17,
User offline. Last seen 42 years 45 weeks ago.
Aw: GitHub move?
Maybe you could write all these problems up and submit them to the GitHub people ... I think they are already aware of the problem. But seeing that some of the bigger, better known projects have problems could make them fix it faster.
Ruediger Keller 2
Joined: 2010-04-30,
User offline. Last seen 42 years 45 weeks ago.
Re: Aw: GitHub move?
Hi Simon,

a little bit off topic, but I also recently had the need to combine several commits into one. What's the easiest way to do that with git?

Regards,
Rüdiger


2011/7/24 Simon Ochsenreither <simon [dot] ochsenreither [at] googlemail [dot] com>
Yes, I have hit some of the same problems. Not sure if there are improvements planned from the GitHub people.

I just want to mention that I think that GitHub adds huge benefits, even without that sort of codereview. I personally handled this problem by closing the pull request, fixing the things mentioned in the comments locally, combining it into a single commit with the older changes and doing a new pull request.

Maxime Lévesque
Joined: 2009-08-18,
User offline. Last seen 42 years 45 weeks ago.
Re: Aw: GitHub move?

If you have a chain of contiguous commits, you do a git reset the oldest commit, then all of your
commits become uncommited modifs, then you can commit them as a whole.
If the commits are non contiguous, then I don't know...

ML

2011/7/24 Ruediger Keller <ruediger [dot] keller [at] rk42 [dot] de>
Hi Simon,

a little bit off topic, but I also recently had the need to combine several commits into one. What's the easiest way to do that with git?

Regards,
Rüdiger


2011/7/24 Simon Ochsenreither <simon [dot] ochsenreither [at] googlemail [dot] com>
Yes, I have hit some of the same problems. Not sure if there are improvements planned from the GitHub people.

I just want to mention that I think that GitHub adds huge benefits, even without that sort of codereview. I personally handled this problem by closing the pull request, fixing the things mentioned in the comments locally, combining it into a single commit with the older changes and doing a new pull request.


gkossakowski
Joined: 2010-03-11,
User offline. Last seen 33 weeks 5 days ago.
Re: Aw: GitHub move?
2011/7/24 Maxime Lévesque <maxime [dot] levesque [at] gmail [dot] com>

If you have a chain of contiguous commits, you do a git reset the oldest commit, then all of your
commits become uncommited modifs, then you can commit them as a whole.
If the commits are non contiguous, then I don't know...

git rebase -i commit_id
and then reorder commits, and mark them with squash command. Here commit_id stands for commit that is a parent of first commit that you want to include in whole process. Read more here: http://kernel.org/pub/software/scm/git/docs/git-rebase.html
However, I'd like to keep this thread on topic, please.

--
Grzegorz Kossakowski

Cédric Beust ♔
Joined: 2011-06-17,
User offline. Last seen 42 years 45 weeks ago.
Re: GitHub move?
Hi Grzegorz,
Indeed, once you've been spoiled by Gerrit, it's hard to go back.
Having said that, maybe you don't have to abandon Gerrit just to move to Github. Gerrit acts as a middle-man between developers and the final repo. Traditionally, this final repo is on the same machine/LAN as Gerrit itself, but theoretically, it should be possible to run Gerrit on a private network somewhere and ask it to push to Github.
The only problem I can see is that you don't have full control over your Github repo, which needs some configurations (creation of specific refs, etc...) in order to work with Github.
Have you asked the Github people what their position is with respect to Gerrit? I wouldn't be surprised if supporting it was in their plans and they might be interested in helping a high profile project.
-- Cédric




On Sun, Jul 24, 2011 at 4:15 AM, Grzegorz Kossakowski <grzegorz [dot] kossakowski [at] gmail [dot] com> wrote:
Hi,
We (Scala+GWT folks) moved to GitHub recently from Gerrit instance maintaned by us (Rob Heittman to be precise). There are number of things which we like about GitHub but there's one very disturbing: code review process feeling broken.
Let me illustrate what's broken about it. See following pull request:
https://github.com/gkossakowski/testcodereview/pull/1
I did a code review and left some inline comments. We see second submission, we open a commit with a file and we see:
https://github.com/gkossakowski/testcodereview/commit/8dff2a1f7f9187aeee36bf556d96c9368a15d35e
There's absolutely nothing indicating that there were comments left on this file so it's really hard to see a context for this change. It means I have to rescan the whole history of pull request and make notes what has been fixed and what not.
Also, issues raised during code review were addressed by introducing new commits and thus polluting the history. It's not a coincidence. If you rebase your changes and push them again to pull request you'll see the history of pull request being completely broken:
https://github.com/gkossakowski/testcodereview/pull/2
Now, those are only basic problems with code review at GitHub. Other problems include complete lack of tracking dependencies between pull requests.
Now, you might be wondering why I'm writing all of this here, on scala-debate. There are two reasons:
  1. I guess there are many people on this list using github a lot so I hope you can point me in right directions how to solve this problem.
  2. If there are no easy solutions to problems I mentioned then I'll be in position to seriously question GitHub move of Scala. I don't question Git move to Scala, but GitHub seems to add no value and may establish false expectations. Namely, people might think it will be easier to contribute to Scala. I cannot see this happen if there are no good code review tools in place.
To sum up: GitHub seems to be very limited when it comes to handling Google-style code review process. If you dig more you'll find even more issues like that. Given how Scala team busy is I'm afraid that GitHub move won't meet high expectations of the Scala community.
PS. These are solely my views on the issue. I didn't discuss this with either Typesafe or Scala team.
-- 
Grzegorz Kossakowski


extempore
Joined: 2008-12-17,
User offline. Last seen 35 weeks 3 days ago.
Re: GitHub move?

I agree github has many issues in this domain. But the move can only
be compared to the alternatives. It cannot possibly be worse than svn
is. If you would like to propose an alternative host, now's the time
to make the case. Otherwise github is the winner by default, and
we'll deal with issues as they arise.

Seth Tisue
Joined: 2008-12-16,
User offline. Last seen 34 weeks 3 days ago.
Re: GitHub move?

On Sun, Jul 24, 2011 at 10:53 AM, Paul Phillips wrote:
> I agree github has many issues in this domain.  But the move can only
> be compared to the alternatives.  It cannot possibly be worse than svn
> is.  If you would like to propose an alternative host, now's the time
> to make the case.  Otherwise github is the winner by default, and
> we'll deal with issues as they arise.

Agree. There is an overwhelming consensus in the Scala community for
GitHub at this point. I've rarely seen so many top developers agree on
something.

Grzegorz, I think your concerns are completely valid, but they aren't
enough that we should "seriously question GitHub move of Scala" — not
even close.

A major appeal of GitHub is that they listen to their users and the
site is still improving at a fast clip. You have to consider the
likely future trajectory as well as the known present. In the case of
GitHub I think the present is already pretty bright, but I think the
future looks even brighter.

Jim Powers
Joined: 2011-01-24,
User offline. Last seen 36 weeks 2 days ago.
Re: GitHub move?
Could the use of Crucible potentially satisfy the code-review problem for Scala? (Is Typesafe and/or EPFL planning on using Crucible?)
I'm certainly all for the move to GitHub.

On Sun, Jul 24, 2011 at 11:54 AM, Seth Tisue <seth [at] tisue [dot] net> wrote:
On Sun, Jul 24, 2011 at 10:53 AM, Paul Phillips <paulp [at] improving [dot] org> wrote:
> I agree github has many issues in this domain.  But the move can only
> be compared to the alternatives.  It cannot possibly be worse than svn
> is.  If you would like to propose an alternative host, now's the time
> to make the case.  Otherwise github is the winner by default, and
> we'll deal with issues as they arise.

Agree. There is an overwhelming consensus in the Scala community for
GitHub at this point. I've rarely seen so many top developers agree on
something.

 Grzegorz, I think your concerns are completely valid, but they aren't
enough that we should "seriously question GitHub move of Scala" — not
even close.

A major appeal of GitHub is that they listen to their users and the
site is still improving at a fast clip. You have to consider the
likely future trajectory as well as the known present. In the case of
GitHub I think the present is already pretty bright, but I think the
future looks even brighter.

--
Seth Tisue | Northwestern University | http://tisue.net
lead developer, NetLogo: http://ccl.northwestern.edu/netlogo/



--
Jim Powers
Kevin Wright 2
Joined: 2010-05-30,
User offline. Last seen 26 weeks 4 days ago.
Re: GitHub move?


On 24 July 2011 17:02, Jim Powers <jim [at] casapowers [dot] com> wrote:
Could the use of Crucible potentially satisfy the code-review problem for Scala? (Is Typesafe and/or EPFL planning on using Crucible?)
 https://codereview.scala-lang.org/fisheye/
 
I'm certainly all for the move to GitHub.

On Sun, Jul 24, 2011 at 11:54 AM, Seth Tisue <seth [at] tisue [dot] net> wrote:
On Sun, Jul 24, 2011 at 10:53 AM, Paul Phillips <paulp [at] improving [dot] org> wrote:
> I agree github has many issues in this domain.  But the move can only
> be compared to the alternatives.  It cannot possibly be worse than svn
> is.  If you would like to propose an alternative host, now's the time
> to make the case.  Otherwise github is the winner by default, and
> we'll deal with issues as they arise.

Agree. There is an overwhelming consensus in the Scala community for
GitHub at this point. I've rarely seen so many top developers agree on
something.

 Grzegorz, I think your concerns are completely valid, but they aren't
enough that we should "seriously question GitHub move of Scala" — not
even close.

A major appeal of GitHub is that they listen to their users and the
site is still improving at a fast clip. You have to consider the
likely future trajectory as well as the known present. In the case of
GitHub I think the present is already pretty bright, but I think the
future looks even brighter.

--
Seth Tisue | Northwestern University | http://tisue.net
lead developer, NetLogo: http://ccl.northwestern.edu/netlogo/



--
Jim Powers

Jim Powers
Joined: 2011-01-24,
User offline. Last seen 36 weeks 2 days ago.
Re: GitHub move?
On Sun, Jul 24, 2011 at 12:15 PM, Kevin Wright <kev [dot] lee [dot] wright [at] gmail [dot] com> wrote:
On 24 July 2011 17:02, Jim Powers <jim [at] casapowers [dot] com> wrote:
Could the use of Crucible potentially satisfy the code-review problem for Scala? (Is Typesafe and/or EPFL planning on using Crucible?)
 https://codereview.scala-lang.org/fisheye/

Awesomage! --
Jim Powers
gkossakowski
Joined: 2010-03-11,
User offline. Last seen 33 weeks 5 days ago.
Re: GitHub move?
On 24 July 2011 16:53, Paul Phillips <paulp [at] improving [dot] org> wrote:
I agree github has many issues in this domain.  But the move can only
be compared to the alternatives.  It cannot possibly be worse than svn
is.

This is a bit apples-to-oranges. I'm all for move to Git, no question about that. 
If you would like to propose an alternative host, now's the time
to make the case.  Otherwise github is the winner by default, and
we'll deal with issues as they arise.

This is though issue and that's why I started this thread. What GitHub does well is UI, focus and general simplicity of the service. It's very easy to setup an account, fork a project and start hacking. I believe this model works fairly well for many open source projects but I don't think this model is very scalable.
The situation we have right now in Scala is that community would like to get more involved into development of Scala and would like to make process of contributing easier. I can get a sense that people associate GitHub move with addressing those needs. However, even if we move to github I don't see how this will improve the situation we have right now that it's only Paul that is taking care of pull requests. The reason is simple: in order to get involved other folks from Scala team in accepting contributions they must see worth it to spend their time on reviewing submitted changes. It's worth it if the amount of time they spend on the process is smaller than their perceived value of the contribution.
One way to address those problem is to have contributions of excellent quality - hard. Another is to reduce time needed to execute the process of code reviewing and generally accepting patches. In my very own opinion, GitHub does not help enough with that. It help with driving people to fork and hack but it does not help to consume those patches and more importantly it does not help enough to ensure that accepted patches are of high quality.
Another solution is use of Gerrit. It's a code review system for Git used by Android project for all of it's development (at least AFAIK). It's definitively scalable tool with excellent support for executing efficient code review process. The biggest problem with Gerrit is that it's definitively less witty than GitHub. It doesn't make barrier of entry very low but focuses on power users instead and does it extremely well in my experience. The problem with Gerrit is that there's no hosting for it apart from Assembla that has it's own set of problems. I cannot see crowds willing to setup and maintain Scala-specific instance of Gerrit.
To sum up: I don't really understand why GitHub move got most votes. I fail to find any document that describes in detail how GitHub will help Scala and how to ensure that imagined process is scalable.
P.S. I'm very biased in my opinions because I happened to work on Gerrit a bit two years ago and I'm have a lot of respect to Shawn Pearce for creating Gerrit. Also, I moved from Gerrit to GitHub and suffer from it. Moreover, as Cedric did point out - you must try Gerrit in practice to understand why I don't get excited about GitHub at all.
P.S. Yep, tomorrow I'll try to contact GitHub folks and hope to have a good conversation with them. If they can solve code review problem I'll fall in love with the service as everyone else here.
--
Grzegorz Kossakowski

Naftoli Gugenheim
Joined: 2008-12-17,
User offline. Last seen 42 years 45 weeks ago.
Re: GitHub move?
An alternative to gerrit is Review Board --- that's what Lift uses.

On Sun, Jul 24, 2011 at 6:15 PM, Grzegorz Kossakowski <grzegorz [dot] kossakowski [at] gmail [dot] com> wrote:
On 24 July 2011 16:53, Paul Phillips <paulp [at] improving [dot] org> wrote:
I agree github has many issues in this domain.  But the move can only
be compared to the alternatives.  It cannot possibly be worse than svn
is.

This is a bit apples-to-oranges. I'm all for move to Git, no question about that. 
If you would like to propose an alternative host, now's the time
to make the case.  Otherwise github is the winner by default, and
we'll deal with issues as they arise.

This is though issue and that's why I started this thread. What GitHub does well is UI, focus and general simplicity of the service. It's very easy to setup an account, fork a project and start hacking. I believe this model works fairly well for many open source projects but I don't think this model is very scalable.
The situation we have right now in Scala is that community would like to get more involved into development of Scala and would like to make process of contributing easier. I can get a sense that people associate GitHub move with addressing those needs. However, even if we move to github I don't see how this will improve the situation we have right now that it's only Paul that is taking care of pull requests. The reason is simple: in order to get involved other folks from Scala team in accepting contributions they must see worth it to spend their time on reviewing submitted changes. It's worth it if the amount of time they spend on the process is smaller than their perceived value of the contribution.
One way to address those problem is to have contributions of excellent quality - hard. Another is to reduce time needed to execute the process of code reviewing and generally accepting patches. In my very own opinion, GitHub does not help enough with that. It help with driving people to fork and hack but it does not help to consume those patches and more importantly it does not help enough to ensure that accepted patches are of high quality.
Another solution is use of Gerrit. It's a code review system for Git used by Android project for all of it's development (at least AFAIK). It's definitively scalable tool with excellent support for executing efficient code review process. The biggest problem with Gerrit is that it's definitively less witty than GitHub. It doesn't make barrier of entry very low but focuses on power users instead and does it extremely well in my experience. The problem with Gerrit is that there's no hosting for it apart from Assembla that has it's own set of problems. I cannot see crowds willing to setup and maintain Scala-specific instance of Gerrit.
To sum up: I don't really understand why GitHub move got most votes. I fail to find any document that describes in detail how GitHub will help Scala and how to ensure that imagined process is scalable.
P.S. I'm very biased in my opinions because I happened to work on Gerrit a bit two years ago and I'm have a lot of respect to Shawn Pearce for creating Gerrit. Also, I moved from Gerrit to GitHub and suffer from it. Moreover, as Cedric did point out - you must try Gerrit in practice to understand why I don't get excited about GitHub at all.
P.S. Yep, tomorrow I'll try to contact GitHub folks and hope to have a good conversation with them. If they can solve code review problem I'll fall in love with the service as everyone else here.
--
Grzegorz Kossakowski


Russ P.
Joined: 2009-01-31,
User offline. Last seen 1 year 26 weeks ago.
Re: GitHub move?
Are you guys aware of Fossil? Yeah, weird name, but it's a great free, distributed version control system. It's "serverless" for ultra simplicity, and it includes a built-in bug tracker and web interface. Also, it's Git compatible, so you can always switch over to Git or any other Git compatible system with little effort if you are unhappy with it. OK, I'm not a power user, so I can only say so much about it, but I think you should take a serious look at it. I bet you will will be impressed.

--Russ P.


On Sun, Jul 24, 2011 at 3:15 PM, Grzegorz Kossakowski <grzegorz [dot] kossakowski [at] gmail [dot] com> wrote:
On 24 July 2011 16:53, Paul Phillips <paulp [at] improving [dot] org> wrote:
I agree github has many issues in this domain.  But the move can only
be compared to the alternatives.  It cannot possibly be worse than svn
is.

This is a bit apples-to-oranges. I'm all for move to Git, no question about that. 
If you would like to propose an alternative host, now's the time
to make the case.  Otherwise github is the winner by default, and
we'll deal with issues as they arise.

This is though issue and that's why I started this thread. What GitHub does well is UI, focus and general simplicity of the service. It's very easy to setup an account, fork a project and start hacking. I believe this model works fairly well for many open source projects but I don't think this model is very scalable.
The situation we have right now in Scala is that community would like to get more involved into development of Scala and would like to make process of contributing easier. I can get a sense that people associate GitHub move with addressing those needs. However, even if we move to github I don't see how this will improve the situation we have right now that it's only Paul that is taking care of pull requests. The reason is simple: in order to get involved other folks from Scala team in accepting contributions they must see worth it to spend their time on reviewing submitted changes. It's worth it if the amount of time they spend on the process is smaller than their perceived value of the contribution.
One way to address those problem is to have contributions of excellent quality - hard. Another is to reduce time needed to execute the process of code reviewing and generally accepting patches. In my very own opinion, GitHub does not help enough with that. It help with driving people to fork and hack but it does not help to consume those patches and more importantly it does not help enough to ensure that accepted patches are of high quality.
Another solution is use of Gerrit. It's a code review system for Git used by Android project for all of it's development (at least AFAIK). It's definitively scalable tool with excellent support for executing efficient code review process. The biggest problem with Gerrit is that it's definitively less witty than GitHub. It doesn't make barrier of entry very low but focuses on power users instead and does it extremely well in my experience. The problem with Gerrit is that there's no hosting for it apart from Assembla that has it's own set of problems. I cannot see crowds willing to setup and maintain Scala-specific instance of Gerrit.
To sum up: I don't really understand why GitHub move got most votes. I fail to find any document that describes in detail how GitHub will help Scala and how to ensure that imagined process is scalable.
P.S. I'm very biased in my opinions because I happened to work on Gerrit a bit two years ago and I'm have a lot of respect to Shawn Pearce for creating Gerrit. Also, I moved from Gerrit to GitHub and suffer from it. Moreover, as Cedric did point out - you must try Gerrit in practice to understand why I don't get excited about GitHub at all.
P.S. Yep, tomorrow I'll try to contact GitHub folks and hope to have a good conversation with them. If they can solve code review problem I'll fall in love with the service as everyone else here.
--
Grzegorz Kossakowski




--
http://RussP.us
extempore
Joined: 2008-12-17,
User offline. Last seen 35 weeks 3 days ago.
Re: GitHub move?

On 7/24/11 3:15 PM, Grzegorz Kossakowski wrote:
> This is a bit apples-to-oranges. I'm all for move to Git, no
> question about that.

If you can only offer assembla as an alternative, then apples vs. apples
is assembla vs. github. Regardless of their relative merits on this
issue, I think if we took a vote github would win by a landslide.

> However, even if we move to github I don't see how this will improve
> the situation we have right now that it's only Paul that is taking
> care of pull requests.

Even paulp is barely doing that, and it's because the overhead to do so
is stratospheric when everything has to be shunted back into svn.

> I fail to find any document that describes in detail how GitHub will
> help Scala and how to ensure that imagined process is scalable.

If that document did exist it would be comparing the current process to
the "imagined process" and would conclude by stating quite convincingly
"the imagined process is much more scalable than the current process."
It doesn't have to be perfect, it only has to be enough better to
warrant the change.

> P.S. I'm very biased in my opinions because I happened to work on
> Gerrit a bit two years ago and I'm have a lot of respect to Shawn
> Pearce for creating Gerrit. Also, I moved from Gerrit to GitHub and
> suffer from it. Moreover, as Cedric did point out - you must try
> Gerrit in practice to understand why I don't get excited about
> GitHub at all.

You sound like you've heard people arguing against gerrit, which I
haven't noticed. Gerrit sounds great to me. Code review is one thing
of many.

> P.S. Yep, tomorrow I'll try to contact GitHub folks and hope to have
> a good conversation with them. If they can solve code review problem
> I'll fall in love with the service as everyone else here.

Put down your strawmen. I have a litany of complaints about github.
I'm also the guy who has absorbed most of the difficulties of
maintaining svn/git-svn/git for 2+ years, and this needs to end. This
conversation is academic unless you can suggest an immediately available
alternative which, considered in total, is more appealing than github.

rytz
Joined: 2008-07-01,
User offline. Last seen 45 weeks 5 days ago.
Re: GitHub move?
We used to have Review Board, and were pretty unhappy with it. Crucible is a bitbetter, but AFAIK some people are not very convinced of this one either..


On Mon, Jul 25, 2011 at 00:26, Naftoli Gugenheim <naftoligug [at] gmail [dot] com> wrote:
An alternative to gerrit is Review Board --- that's what Lift uses.

On Sun, Jul 24, 2011 at 6:15 PM, Grzegorz Kossakowski <grzegorz [dot] kossakowski [at] gmail [dot] com> wrote:
On 24 July 2011 16:53, Paul Phillips <paulp [at] improving [dot] org> wrote:
I agree github has many issues in this domain.  But the move can only
be compared to the alternatives.  It cannot possibly be worse than svn
is.

This is a bit apples-to-oranges. I'm all for move to Git, no question about that. 
If you would like to propose an alternative host, now's the time
to make the case.  Otherwise github is the winner by default, and
we'll deal with issues as they arise.

This is though issue and that's why I started this thread. What GitHub does well is UI, focus and general simplicity of the service. It's very easy to setup an account, fork a project and start hacking. I believe this model works fairly well for many open source projects but I don't think this model is very scalable.
The situation we have right now in Scala is that community would like to get more involved into development of Scala and would like to make process of contributing easier. I can get a sense that people associate GitHub move with addressing those needs. However, even if we move to github I don't see how this will improve the situation we have right now that it's only Paul that is taking care of pull requests. The reason is simple: in order to get involved other folks from Scala team in accepting contributions they must see worth it to spend their time on reviewing submitted changes. It's worth it if the amount of time they spend on the process is smaller than their perceived value of the contribution.
One way to address those problem is to have contributions of excellent quality - hard. Another is to reduce time needed to execute the process of code reviewing and generally accepting patches. In my very own opinion, GitHub does not help enough with that. It help with driving people to fork and hack but it does not help to consume those patches and more importantly it does not help enough to ensure that accepted patches are of high quality.
Another solution is use of Gerrit. It's a code review system for Git used by Android project for all of it's development (at least AFAIK). It's definitively scalable tool with excellent support for executing efficient code review process. The biggest problem with Gerrit is that it's definitively less witty than GitHub. It doesn't make barrier of entry very low but focuses on power users instead and does it extremely well in my experience. The problem with Gerrit is that there's no hosting for it apart from Assembla that has it's own set of problems. I cannot see crowds willing to setup and maintain Scala-specific instance of Gerrit.
To sum up: I don't really understand why GitHub move got most votes. I fail to find any document that describes in detail how GitHub will help Scala and how to ensure that imagined process is scalable.
P.S. I'm very biased in my opinions because I happened to work on Gerrit a bit two years ago and I'm have a lot of respect to Shawn Pearce for creating Gerrit. Also, I moved from Gerrit to GitHub and suffer from it. Moreover, as Cedric did point out - you must try Gerrit in practice to understand why I don't get excited about GitHub at all.
P.S. Yep, tomorrow I'll try to contact GitHub folks and hope to have a good conversation with them. If they can solve code review problem I'll fall in love with the service as everyone else here.
--
Grzegorz Kossakowski



Naftoli Gugenheim
Joined: 2008-12-17,
User offline. Last seen 42 years 45 weeks ago.
Re: GitHub move?
I'm curious what its downsides are (obviously as compared to other products, since I've only used RB).

On Mon, Jul 25, 2011 at 2:46 AM, Lukas Rytz <lukas [dot] rytz [at] epfl [dot] ch> wrote:
We used to have Review Board, and were pretty unhappy with it. Crucible is a bitbetter, but AFAIK some people are not very convinced of this one either..


On Mon, Jul 25, 2011 at 00:26, Naftoli Gugenheim <naftoligug [at] gmail [dot] com> wrote:
An alternative to gerrit is Review Board --- that's what Lift uses.

On Sun, Jul 24, 2011 at 6:15 PM, Grzegorz Kossakowski <grzegorz [dot] kossakowski [at] gmail [dot] com> wrote:
On 24 July 2011 16:53, Paul Phillips <paulp [at] improving [dot] org> wrote:
I agree github has many issues in this domain.  But the move can only
be compared to the alternatives.  It cannot possibly be worse than svn
is.

This is a bit apples-to-oranges. I'm all for move to Git, no question about that. 
If you would like to propose an alternative host, now's the time
to make the case.  Otherwise github is the winner by default, and
we'll deal with issues as they arise.

This is though issue and that's why I started this thread. What GitHub does well is UI, focus and general simplicity of the service. It's very easy to setup an account, fork a project and start hacking. I believe this model works fairly well for many open source projects but I don't think this model is very scalable.
The situation we have right now in Scala is that community would like to get more involved into development of Scala and would like to make process of contributing easier. I can get a sense that people associate GitHub move with addressing those needs. However, even if we move to github I don't see how this will improve the situation we have right now that it's only Paul that is taking care of pull requests. The reason is simple: in order to get involved other folks from Scala team in accepting contributions they must see worth it to spend their time on reviewing submitted changes. It's worth it if the amount of time they spend on the process is smaller than their perceived value of the contribution.
One way to address those problem is to have contributions of excellent quality - hard. Another is to reduce time needed to execute the process of code reviewing and generally accepting patches. In my very own opinion, GitHub does not help enough with that. It help with driving people to fork and hack but it does not help to consume those patches and more importantly it does not help enough to ensure that accepted patches are of high quality.
Another solution is use of Gerrit. It's a code review system for Git used by Android project for all of it's development (at least AFAIK). It's definitively scalable tool with excellent support for executing efficient code review process. The biggest problem with Gerrit is that it's definitively less witty than GitHub. It doesn't make barrier of entry very low but focuses on power users instead and does it extremely well in my experience. The problem with Gerrit is that there's no hosting for it apart from Assembla that has it's own set of problems. I cannot see crowds willing to setup and maintain Scala-specific instance of Gerrit.
To sum up: I don't really understand why GitHub move got most votes. I fail to find any document that describes in detail how GitHub will help Scala and how to ensure that imagined process is scalable.
P.S. I'm very biased in my opinions because I happened to work on Gerrit a bit two years ago and I'm have a lot of respect to Shawn Pearce for creating Gerrit. Also, I moved from Gerrit to GitHub and suffer from it. Moreover, as Cedric did point out - you must try Gerrit in practice to understand why I don't get excited about GitHub at all.
P.S. Yep, tomorrow I'll try to contact GitHub folks and hope to have a good conversation with them. If they can solve code review problem I'll fall in love with the service as everyone else here.
--
Grzegorz Kossakowski




rytz
Joined: 2008-07-01,
User offline. Last seen 45 weeks 5 days ago.
Re: GitHub move?
I just wanted to state our history of review tools. I cannot say what individualsdon't / didn't like about them.
On Mon, Jul 25, 2011 at 08:49, Naftoli Gugenheim <naftoligug [at] gmail [dot] com> wrote:
I'm curious what its downsides are (obviously as compared to other products, since I've only used RB).

On Mon, Jul 25, 2011 at 2:46 AM, Lukas Rytz <lukas [dot] rytz [at] epfl [dot] ch> wrote:
We used to have Review Board, and were pretty unhappy with it. Crucible is a bitbetter, but AFAIK some people are not very convinced of this one either..


On Mon, Jul 25, 2011 at 00:26, Naftoli Gugenheim <naftoligug [at] gmail [dot] com> wrote:
An alternative to gerrit is Review Board --- that's what Lift uses.

On Sun, Jul 24, 2011 at 6:15 PM, Grzegorz Kossakowski <grzegorz [dot] kossakowski [at] gmail [dot] com> wrote:
On 24 July 2011 16:53, Paul Phillips <paulp [at] improving [dot] org> wrote:
I agree github has many issues in this domain.  But the move can only
be compared to the alternatives.  It cannot possibly be worse than svn
is.

This is a bit apples-to-oranges. I'm all for move to Git, no question about that. 
If you would like to propose an alternative host, now's the time
to make the case.  Otherwise github is the winner by default, and
we'll deal with issues as they arise.

This is though issue and that's why I started this thread. What GitHub does well is UI, focus and general simplicity of the service. It's very easy to setup an account, fork a project and start hacking. I believe this model works fairly well for many open source projects but I don't think this model is very scalable.
The situation we have right now in Scala is that community would like to get more involved into development of Scala and would like to make process of contributing easier. I can get a sense that people associate GitHub move with addressing those needs. However, even if we move to github I don't see how this will improve the situation we have right now that it's only Paul that is taking care of pull requests. The reason is simple: in order to get involved other folks from Scala team in accepting contributions they must see worth it to spend their time on reviewing submitted changes. It's worth it if the amount of time they spend on the process is smaller than their perceived value of the contribution.
One way to address those problem is to have contributions of excellent quality - hard. Another is to reduce time needed to execute the process of code reviewing and generally accepting patches. In my very own opinion, GitHub does not help enough with that. It help with driving people to fork and hack but it does not help to consume those patches and more importantly it does not help enough to ensure that accepted patches are of high quality.
Another solution is use of Gerrit. It's a code review system for Git used by Android project for all of it's development (at least AFAIK). It's definitively scalable tool with excellent support for executing efficient code review process. The biggest problem with Gerrit is that it's definitively less witty than GitHub. It doesn't make barrier of entry very low but focuses on power users instead and does it extremely well in my experience. The problem with Gerrit is that there's no hosting for it apart from Assembla that has it's own set of problems. I cannot see crowds willing to setup and maintain Scala-specific instance of Gerrit.
To sum up: I don't really understand why GitHub move got most votes. I fail to find any document that describes in detail how GitHub will help Scala and how to ensure that imagined process is scalable.
P.S. I'm very biased in my opinions because I happened to work on Gerrit a bit two years ago and I'm have a lot of respect to Shawn Pearce for creating Gerrit. Also, I moved from Gerrit to GitHub and suffer from it. Moreover, as Cedric did point out - you must try Gerrit in practice to understand why I don't get excited about GitHub at all.
P.S. Yep, tomorrow I'll try to contact GitHub folks and hope to have a good conversation with them. If they can solve code review problem I'll fall in love with the service as everyone else here.
--
Grzegorz Kossakowski





Kevin Wright 2
Joined: 2010-05-30,
User offline. Last seen 26 weeks 4 days ago.
Re: GitHub move?


On 24 July 2011 23:15, Grzegorz Kossakowski <grzegorz [dot] kossakowski [at] gmail [dot] com> wrote:
On 24 July 2011 16:53, Paul Phillips <paulp [at] improving [dot] org> wrote:
I agree github has many issues in this domain.  But the move can only
be compared to the alternatives.  It cannot possibly be worse than svn
is.

This is a bit apples-to-oranges. I'm all for move to Git, no question about that.

This, I think, is the entire point... We're moving to Git 
GitHub is incidental to this, it just happens to be the most popular choice of hosting environment for our chosen DVCS.(and I must emphasis that it's distributed, this means that moving from GitHub to another environment is as easy as cloning any existing repo, full history comes for free.)
For now, issue tracking and code reviews are handled by Jira and Crucible, with a Confluence wiki as well.  This means that we needn't worry about Gerrit vs GitHub vs Whatever as this stage with regards to such tools.  Maybe we *will* migrate code reviews or issue tracking at some point in the future, but I personally feel that it would be risky to do so at the same time as migrating version control (unless there were no other choice)
About the only feature from GitHub we need right now is pull requests.  Combine that with freedom from git-svn integration and we're good to go.  We can then migrate to the linux-esque distributed model for which Git was originally designed: a number Downstream repos would accept patches, each in a specific area (e.g. documentation, collections, io, compiler, swing, etc.), with a nominated person then accepting pull requests for each of these sub-repos before pushing an aggregated changeset up to the "central" repository for final aggregation by Paul.
Paul should then (in theory) have fewer pull requests and some more assurance that they apply cleanly, thus allowing him to grow back his hair without worrying that he'll pull it out again.  We'll also all have more time to hack some extra awesomeness into the language.
gkossakowski
Joined: 2010-03-11,
User offline. Last seen 33 weeks 5 days ago.
Re: GitHub move?
On 25 July 2011 01:07, Paul Phillips <paulp [at] improving [dot] org> wrote:
On 7/24/11 3:15 PM, Grzegorz Kossakowski wrote:
This is a bit apples-to-oranges. I'm all for move to Git, no
question about that.

If you can only offer assembla as an alternative, then apples vs. apples
is assembla vs. github.  Regardless of their relative merits on this
issue, I think if we took a vote github would win by a landslide.

I assume that we keep our wikis, issue tracker, etc. intact. Then we need only a hosting for repositories + code review, right?
If my assumption is right, then I'd propose Gerrit instance hosted by EPFL instead of Assembla. It's way easier to customize Gerrit if we control it.
I guess that Gerrit would not get many votes too.
I fail to find any document that describes in detail how GitHub will
help Scala and how to ensure that imagined process is scalable.

If that document did exist it would be comparing the current process to the "imagined process" and would conclude by stating quite convincingly "the imagined process is much more scalable than the current process." It doesn't have to be perfect, it only has to be enough better to warrant the change.

That's probably main point where we disagree. I see Git as a major improvement to warrant the change, GitHub not so. 


P.S. Yep, tomorrow I'll try to contact GitHub folks and hope to have
a good conversation with them. If they can solve code review problem
I'll fall in love with the service as everyone else here.

Put down your strawmen.  I have a litany of complaints about github. I'm also the guy who has absorbed most of the difficulties of maintaining svn/git-svn/git for 2+ years, and this needs to end.  This conversation is academic unless you can suggest an immediately available alternative which, considered in total, is more appealing than github.

I think we reached a point where it's hard to have a productive discussion. I'd need to better understand which features of GitHub are planned to be used. I'd need to know if we keep JIRA as issue tracker and Confluence as wiki. I'd need to know if all development is planned to be done on GitHub or only community part.
I think I should make everyone a favor and explain why I started this thread. Two reasons:
  • to vent off a bit of my frustration related to GitHub and maybe spur people to propose some solutions to my problems. I honestly thought I'm missing something obvious to GitHubbers which seems to not be the case
  • I wanted to let people know that GitHub move might not come up all roses.

--
Grzegorz Kossakowski

Kevin Wright 2
Joined: 2010-05-30,
User offline. Last seen 26 weeks 4 days ago.
Re: GitHub move?

I think we reached a point where it's hard to have a productive discussion. I'd need to better understand which features of GitHub are planned to be used. I'd need to know if we keep JIRA as issue tracker and Confluence as wiki. I'd need to know if all development is planned to be done on GitHub or only community part.

GIt doesn't work like that.  Development is done *on* your local repository, held in the .git directory at the root of the source tree.
There can then be one or more shared repositories used for integrating changes from multiple people, GitHub is primarily a tool for simplifying this process. The idea of some central repository is then nothing more than convention, though it *is* normal to have such a concept - because, amongst other things, this gives an obvious place to integrate additional tooling.
gkossakowski
Joined: 2010-03-11,
User offline. Last seen 33 weeks 5 days ago.
Re: GitHub move?
On 25 July 2011 10:30, Kevin Wright <kev [dot] lee [dot] wright [at] gmail [dot] com> wrote:

I think we reached a point where it's hard to have a productive discussion. I'd need to better understand which features of GitHub are planned to be used. I'd need to know if we keep JIRA as issue tracker and Confluence as wiki. I'd need to know if all development is planned to be done on GitHub or only community part.

GIt doesn't work like that.  Development is done *on* your local repository, held in the .git directory at the root of the source tree.
There can then be one or more shared repositories used for integrating changes from multiple people, GitHub is primarily a tool for simplifying this process. The idea of some central repository is then nothing more than convention, though it *is* normal to have such a concept - because, amongst other things, this gives an obvious place to integrate additional tooling.

I guess I wasn't clear what I was asking: if we move to GitHub, are all changes going to be reviewed by issuing pull requests, including changes by core Scala team?
I should mention that I understand Git model fairly well, so we can skip that part :-)

--
Grzegorz Kossakowski

rytz
Joined: 2008-07-01,
User offline. Last seen 45 weeks 5 days ago.
Re: GitHub move?


On Mon, Jul 25, 2011 at 10:41, Grzegorz Kossakowski <grzegorz [dot] kossakowski [at] gmail [dot] com> wrote:
On 25 July 2011 10:30, Kevin Wright <kev [dot] lee [dot] wright [at] gmail [dot] com> wrote:

I think we reached a point where it's hard to have a productive discussion. I'd need to better understand which features of GitHub are planned to be used. I'd need to know if we keep JIRA as issue tracker and Confluence as wiki. I'd need to know if all development is planned to be done on GitHub or only community part.

GIt doesn't work like that.  Development is done *on* your local repository, held in the .git directory at the root of the source tree.
There can then be one or more shared repositories used for integrating changes from multiple people, GitHub is primarily a tool for simplifying this process. The idea of some central repository is then nothing more than convention, though it *is* normal to have such a concept - because, amongst other things, this gives an obvious place to integrate additional tooling.

I guess I wasn't clear what I was asking: if we move to GitHub, are all changes going to be reviewed by issuing pull requests, including changes by core Scala team?

No, we keep post-commit reviews for the core committers. 

I should mention that I understand Git model fairly well, so we can skip that part :-)

--
Grzegorz Kossakowski


gkossakowski
Joined: 2010-03-11,
User offline. Last seen 33 weeks 5 days ago.
Re: GitHub move?
On 25 July 2011 10:53, Lukas Rytz <lukas [dot] rytz [at] epfl [dot] ch> wrote:
No, we keep post-commit reviews for the core committers.

Using Crucidble?

--
Grzegorz Kossakowski

Johannes Rudolph 2
Joined: 2010-02-12,
User offline. Last seen 42 years 45 weeks ago.
Re: GitHub move?

I think there are two aspects for the "move to github": The first one
is the move to git, which is basically a technical one. It is mainly
undisputed because it will simplify development for most parties (and
particularly for Paul). But it is also an enabler for the second step,
namely how the social interactions around contribution and code change
will be organized.

And IMO it makes perfect sense to start now to think about what good
models of accepting contributions/patches to adopt for a sizeable
project that Scala has become. Working towards some policy regarding
contribution contacts, process and software support can only be a
benefit for further improvement of the Scala code base. This don't
have to be and shouldn't be a distraction for the technical move to
git, though.

On Mon, Jul 25, 2011 at 10:41 AM, Grzegorz Kossakowski
wrote:
> On 25 July 2011 10:30, Kevin Wright wrote:
>>>
>>> I think we reached a point where it's hard to have a productive
>>> discussion. I'd need to better understand which features of GitHub are
>>> planned to be used. I'd need to know if we keep JIRA as issue tracker and
>>> Confluence as wiki. I'd need to know if all development is planned to be
>>> done on GitHub or only community part.
>>
>> GIt doesn't work like that.  Development is done *on* your local
>> repository, held in the .git directory at the root of the source tree.
>> There can then be one or more shared repositories used for integrating
>> changes from multiple people, GitHub is primarily a tool for simplifying
>> this process. The idea of some central repository is then nothing more than
>> convention, though it *is* normal to have such a concept - because, amongst
>> other things, this gives an obvious place to integrate additional tooling.
>
> I guess I wasn't clear what I was asking: if we move to GitHub, are all
> changes going to be reviewed by issuing pull requests, including changes by
> core Scala team?
> I should mention that I understand Git model fairly well, so we can skip
> that part :-)
>
> --
> Grzegorz Kossakowski
>
>

rytz
Joined: 2008-07-01,
User offline. Last seen 45 weeks 5 days ago.
Re: GitHub move?


On Mon, Jul 25, 2011 at 10:56, Grzegorz Kossakowski <grzegorz [dot] kossakowski [at] gmail [dot] com> wrote:
On 25 July 2011 10:53, Lukas Rytz <lukas [dot] rytz [at] epfl [dot] ch> wrote:
No, we keep post-commit reviews for the core committers.

Using Crucidble?

yes, for the moment there's no initiative to replace crucible.
Joshua.Suereth
Joined: 2008-09-02,
User offline. Last seen 32 weeks 5 days ago.
Re: GitHub move?
So how about we take this approach.
(1) Move Scala to git.   This requires solving a few technical issues, and I'm tackling this ASAP.  In particular, STARR will no longer be in the source repository, but a binary repostiory (either Ivy or Maven).
(2) For now, we use github because it's where folks know to look for code, and crucible because it's in use today.    After the move to github, since git is distributed, we have some flexibility in choices, unless Gerrit is locked down in some strange manner that would prevent migrating an existing git repo to it. 
(3)  In a month or two's time, if github proves to be a burden, we can set up a Gerrit instance to try out.   We can still attempt to use github either as a public endpoint of all things scala, *or* we could funnel github 'development' branch through gerrit and then back into github under 'master'.
Let's chat a bit when we have time.  
- Josh
On Mon, Jul 25, 2011 at 5:03 AM, Johannes Rudolph <johannes [dot] rudolph [at] googlemail [dot] com> wrote:
I think there are two aspects for the "move to github": The first one
is the move to git, which is basically a technical one. It is mainly
undisputed because it will simplify development for most parties (and
particularly for Paul). But it is also an enabler for the second step,
namely how the social interactions around contribution and code change
will be organized.

And IMO it makes perfect sense to start now to think about what good
models of accepting contributions/patches to adopt for a sizeable
project that Scala has become. Working towards some policy regarding
contribution contacts, process and software support can only be a
benefit for further improvement of the Scala code base. This don't
have to be and shouldn't be a distraction for the technical move to
git, though.


On Mon, Jul 25, 2011 at 10:41 AM, Grzegorz Kossakowski
<grzegorz [dot] kossakowski [at] gmail [dot] com> wrote:
> On 25 July 2011 10:30, Kevin Wright <kev [dot] lee [dot] wright [at] gmail [dot] com> wrote:
>>>
>>> I think we reached a point where it's hard to have a productive
>>> discussion. I'd need to better understand which features of GitHub are
>>> planned to be used. I'd need to know if we keep JIRA as issue tracker and
>>> Confluence as wiki. I'd need to know if all development is planned to be
>>> done on GitHub or only community part.
>>
>> GIt doesn't work like that.  Development is done *on* your local
>> repository, held in the .git directory at the root of the source tree.
>> There can then be one or more shared repositories used for integrating
>> changes from multiple people, GitHub is primarily a tool for simplifying
>> this process. The idea of some central repository is then nothing more than
>> convention, though it *is* normal to have such a concept - because, amongst
>> other things, this gives an obvious place to integrate additional tooling.
>
> I guess I wasn't clear what I was asking: if we move to GitHub, are all
> changes going to be reviewed by issuing pull requests, including changes by
> core Scala team?
> I should mention that I understand Git model fairly well, so we can skip
> that part :-)
>
> --
> Grzegorz Kossakowski
>
>



--
Johannes

-----------------------------------------------
Johannes Rudolph
http://virtual-void.net

gkossakowski
Joined: 2010-03-11,
User offline. Last seen 33 weeks 5 days ago.
Re: GitHub move?
On 25 July 2011 15:07, Josh Suereth <joshua [dot] suereth [at] gmail [dot] com> wrote:
So how about we take this approach.
(1) Move Scala to git.   This requires solving a few technical issues, and I'm tackling this ASAP.  In particular, STARR will no longer be in the source repository, but a binary repostiory (either Ivy or Maven).
(2) For now, we use github because it's where folks know to look for code, and crucible because it's in use today.    After the move to github, since git is distributed, we have some flexibility in choices, unless Gerrit is locked down in some strange manner that would prevent migrating an existing git repo to it. 
(3)  In a month or two's time, if github proves to be a burden, we can set up a Gerrit instance to try out.   We can still attempt to use github either as a public endpoint of all things scala, *or* we could funnel github 'development' branch through gerrit and then back into github under 'master'.

LGTM.
--
Grzegorz Kossakowski

Simon Ochsenreither
Joined: 2011-07-17,
User offline. Last seen 42 years 45 weeks ago.
Aw: Re: GitHub move?
Hi,

The reason is simple: in order to get involved other folks from Scala team in accepting contributions they must see worth it to spend their time on reviewing submitted changes. It's worth it if the amount of time they spend on the process is smaller than their perceived value of the contribution.
Agree!
 
One way to address those problem is to have contributions of excellent quality - hard.
 
Another is to reduce time needed to execute the process of code reviewing and generally accepting patches.
 
Another solution is use of Gerrit.

I think you are missing one solution:

Adding more people who are able to merge pull requests. Do we really need to force a compiler expert to review trivial enhancements to ScalaDoc for instance?
Requiring that contributions need to have excellent quality without any sort of mentoring (to get new people up to speed) is just not practical. New contributors who will instantly deliver perfect code don't really grow on trees these days. The current process is frustrating for new contributors and turns interested people away quickly.

Let's face it: There aren't two dozen new contributors wanting to merge their pattern matching rewrite. It is mostly about people wanting to contribute some documentation, wanting to fix some typo, ... why isn't possible to let someone handle this who "has nothing better to do" instead of choosing someone who has a huge workload?

Imagine this IRC conversation:
19:45 || someNewUser: I found a typo in FooClass! Shouldn't it read "bar" instead of "baz"?
19:46 || guyWithCommitRights: Yes, you are right. Thanks!
19:50 || guyWithCommitRights: I just fixed it. The fix should appear tomorrow in the nightly ScalaDoc.

 - OR -

20:20 || newContributor: I added some documentation to class Baz. Pull request #42. Could you have a look at it?
20:24 || guyWithCommitRights: Looks good. I merged your pull request. Thanks!

IMHO this is how things should work and we should investigate what's necessary to make that happen.
 
To sum up: I don't really understand why GitHub move got most votes. I fail to find any document that describes in detail how GitHub will help Scala and how to ensure that imagined process is scalable.
 
Just to add some perspective: I don't care about Git/Hg/Bazaar/whatever, but GitHub is _the_ killer feature.

Bye
extempore
Joined: 2008-12-17,
User offline. Last seen 35 weeks 3 days ago.
Re: GitHub move?

On 7/25/11 6:09 AM, Grzegorz Kossakowski wrote:
> (2) For now, we use github because it's where folks know to look for
> code, and crucible because it's in use today. After the move to
> github, since git is distributed, we have some flexibility in
> choices, unless Gerrit is locked down in some strange manner that
> would prevent migrating an existing git repo to it.
>
> (3) In a month or two's time, if github proves to be a burden, we
> can set up a Gerrit instance to try out. We can still attempt to
> use github either as a public endpoint of all things scala, *or* we
> could funnel github 'development' branch through gerrit and then back
> into github under 'master'.
>
> LGTM.

Sounds like we're all in agreement. Josh was sensible to articulate it,
but I should point out that I thought these points were givens. My only
interest is results, and nobody is likely to feel inadequacies of the
development process any more keenly than I. We should always be
vigilant about improving that which can be improved. We're not getting
married, and github doesn't have to be "Mrs. Right": she is however
"Mrs. Right Now."

Florian Hars 3
Joined: 2011-05-08,
User offline. Last seen 42 years 45 weeks ago.
Re: GitHub move?

On Sun, Jul 24, 2011 at 03:32:49PM -0700, Russ Paielli wrote:
> Are you guys aware of Fossil? Yeah, weird name, but it's a great free,
> distributed version control system.

And is reported to destroy repositories if someone branches:
http://sheddingbikes.com/posts/1306005291.html

- Florian.

Russ P.
Joined: 2009-01-31,
User offline. Last seen 1 year 26 weeks ago.
Re: GitHub move?
On Mon, Jul 25, 2011 at 10:50 PM, Florian Hars <florian [at] hars [dot] de> wrote:
On Sun, Jul 24, 2011 at 03:32:49PM -0700, Russ Paielli wrote:
> Are you guys aware of Fossil? Yeah, weird name, but it's a great free,
> distributed version control system.

And is reported to destroy repositories if someone branches:
http://sheddingbikes.com/posts/1306005291.html

- Florian.


Interesting. I posted this on the Fossil mailing list, and I got the following reply. I will keep using Fossil and hope that the bug's been fixed. It's just too elegant and minimal to *not* use. As I said before, I am a compulsive minimalist. Here is the reply:

Already discussed at length on the list
(fossil-users [at] lists [dot] fossil-scm [dot] org/msg04665.html" target="_blank" rel="nofollow">http://www.mail-archive.com/fossil-users [at] lists [dot] fossil-scm [dot] org/msg04665.html).

IIRC, there was a subtle bug that caused a "fossil update" to update
to an empty branch - which then removed most of his files (bug fixed
during this discussion). At this point, none of his work was
lost. However, he then panicked (not unreasonable) and in trying to
get things fixed managed to do things that did lose work. I don't
think enough information was ever posted to decide if fossil actually
lost work, or if he just managed to destroy it while trying to recover
from the checkout of nothing.

      <mike
--
Mike Meyer <mwm [at] mired [dot] org>              http://www.mired.org/
Independent Software developer/SCM consultant, email for more information.


--
http://RussP.us

Copyright © 2012 École Polytechnique Fédérale de Lausanne (EPFL), Lausanne, Switzerland