I have had some unpleasant issues with darcs working on cl-json, and I find that working with it is very badly hampered by its lack of mindshare compared with other DVCSes. In particular:
1. The darcs maintainers are unable to supply binaries for Mac OSX. 2. No one has stepped up to the plate to provide a graphical UI for working with darcs.
Not only are these a nuisance in and of themselves (the former especially, since darcs does not build on an out-of-the-box install of the Haskell Platform), they are the signs of a tool that does not have an adequately robust support community.
Combine this with the oddity of the usage model, compared with other DCVSes,[1] and you have real problems.
I am not volunteering to maintain CL-JSON, so it is ultimately not up to me. But I have submitted a number of patches, so I feel that I have standing at least to offer the suggestion.
Note that I don't think that darcs, in and of itself, is a bad system. But there are more than enough DCVSes out there, and we can use one less of them.
Best regards, Robert
[1] See, for example: http://darcs.net/FAQ#how-do-i-go-back-to-an-older-version-of-my-repository
On Thu, Nov 8, 2012 at 4:52 AM, Robert Goldman rpgoldman@sift.info wrote:
Note that I don't think that darcs, in and of itself, is a bad system. But there are more than enough DCVSes out there, and we can use one less of them.
I think it is a good idea, it crossed my mind too. Also github is a good environment for cooperating. I think the darcs repo should live for some time while we double-commit. If no one objects, I'll try to do it quite soon.
/Henrik
On 11/8/12 Nov 8 -1:31 AM, Henrik Hjelte wrote:
On Thu, Nov 8, 2012 at 4:52 AM, Robert Goldman <rpgoldman@sift.info mailto:rpgoldman@sift.info> wrote:
Note that I don't think that darcs, in and of itself, is a bad system. But there are more than enough DCVSes out there, and we can use one less of them.I think it is a good idea, it crossed my mind too. Also github is a good environment for cooperating. I think the darcs repo should live for some time while we double-commit. If no one objects, I'll try to do it quite soon.
Sounds good. I have a preference for a canonical git repo at c-l.net, although a pointer from the c-l.net web page to one on github would be ok. What I'd like us to avoid is that mess where there are a dozen github repos for the same library, and no one knows which one is canonical. And I think we should still provide tarballs.
I don't really get github's additional structure on top of vanilla git usage, and am not that excited about learning it. But I would be happy to continue to submit patches in the old school way, through 'git email'.
Cheers, r
On Thu, Nov 8, 2012 at 4:53 PM, Robert Goldman rpgoldman@sift.info wrote:
On 11/8/12 Nov 8 -1:31 AM, Henrik Hjelte wrote:
On Thu, Nov 8, 2012 at 4:52 AM, Robert Goldman <rpgoldman@sift.info mailto:rpgoldman@sift.info> wrote:
Note that I don't think that darcs, in and of itself, is a badsystem.
But there are more than enough DCVSes out there, and we can use oneless
of them.I think it is a good idea, it crossed my mind too. Also github is a good environment for cooperating. I think the darcs repo should live for some time while we double-commit. If no one objects, I'll try to do it quite soon.
Sounds good. I have a preference for a canonical git repo at c-l.net, although a pointer from the c-l.net web page to one on github would be ok.
I read this mail after I did the git migration, but I did it like that. The docs are still on common-lisp but mentions the github url.
What I'd like us to avoid is that mess where there are a dozen github repos for the same library, and no one knows which one is canonical.
I did a git-repo here: https://github.com/hankhero/cl-json I am not sure it would add something to host it on common-lisp.net instead. Other such as hunchentoot seem to be on github.
And I think we should still provide tarballs.
If someone asks on the mailing-list or me directly for a tarball, then I can do them. But does anyone use them?
I don't really get github's additional structure on top of vanilla git
usage, and am not that excited about learning it.
I think it will stimulate more patches, even from people that don't take their time or wish to send patches for various reasons. I think a lot of people think it is easier to ask for a pull-request. It is just a guess though.
But I would be happy to continue to submit patches in the old school way, through 'git email'.
Well that's great!
Best wishes, Henrik
On 11/12/12 Nov 12 -5:32 PM, Henrik Hjelte wrote:
On Thu, Nov 8, 2012 at 4:53 PM, Robert Goldman <rpgoldman@sift.info mailto:rpgoldman@sift.info> wrote:
On 11/8/12 Nov 8 -1:31 AM, Henrik Hjelte wrote: > On Thu, Nov 8, 2012 at 4:52 AM, Robert Goldman <rpgoldman@sift.info <mailto:rpgoldman@sift.info> > <mailto:rpgoldman@sift.info <mailto:rpgoldman@sift.info>>> wrote: > > Note that I don't think that darcs, in and of itself, is a bad system. > But there are more than enough DCVSes out there, and we can use one less > of them. > > > I think it is a good idea, it crossed my mind too. Also github is a good > environment for cooperating. > I think the darcs repo should live for some time while we double-commit. > If no one objects, I'll try to do it quite soon. Sounds good. I have a preference for a canonical git repo at c-l.net <http://c-l.net>, although a pointer from the c-l.net <http://c-l.net> web page to one on github would be ok.I read this mail after I did the git migration, but I did it like that. The docs are still on common-lisp but mentions the github url.
What I'd like us to avoid is that mess where there are a dozen github repos for the same library, and no one knows which one is canonical.I did a git-repo here: https://github.com/hankhero/cl-json I am not sure it would add something to host it on common-lisp.net http://common-lisp.net instead. Other such as hunchentoot seem to be on github.
That looks great, thanks!
Really appreciate your doing this.
Best, Robert
-
Anton Vodonosov -
Henrik Hjelte -
Robert Goldman