Could you give CVS write permissions for SLIME to the following people?
Jeffrey Cunningham jeffrey@cunningham.net Attila Lendvai attila.lendvai@gmail.com
thanks Helmut!
as you seems to be the long missed dictator (or even if you are not, you have been stamped by this :) what do you and the others think about (1) moving the darcs repo at http://common-lisp.net/cgi-bin/darcsweb/darcsweb.cgi?r=cl-wdim-slime;a=summa... under /project/slime? and (2) advertising it as the primary repository?
sorry for pushing it that much, but i honestly think that it would make everyone's life (including me) easier on the long run.
in case there are people who strongly disagree with this, please speak up and tell your reasons. if it turned out to be desired (which may have been signaled by a long enough silence) i'd voulenteer to do the job.
* Attila Lendvai [2006-11-19 21:19+0100] writes:
as you seems to be the long missed dictator (or even if you are not, you have been stamped by this :)
There's no dictator. I think every Slime hacker has equal rights and responsibilities. This includes giving commit rights and rewriting/deleting code that someone else wrote.
what do you and the others think about (1) moving the darcs repo at http://common-lisp.net/cgi-bin/darcsweb/darcsweb.cgi?r=cl-wdim-slime;a=summa... under /project/slime? and (2) advertising it as the primary repository?
I have no objections against moving to darcs. But that repo only contains the commit history back to 2005-04-04. It wouldn't be wise to throw away earlier changes. I've run cvs2darcs and placed the result at http://common-lisp.net/~heller/slime-cvs2darcs.tar.bz2. It's twice as big: 17MB uncompressed. Is that a problem?
Helmut.
as you seems to be the long missed dictator (or even if you are not, you have been stamped by this :)
There's no dictator. I think every Slime hacker has equal rights and responsibilities. This includes giving commit rights and rewriting/deleting code that someone else wrote.
i was more-or-less joking there... :) i agree with you in these.
I have no objections against moving to darcs. But that repo only contains the commit history back to 2005-04-04. It wouldn't be wise to throw away earlier changes. I've run cvs2darcs and placed the result at http://common-lisp.net/~heller/slime-cvs2darcs.tar.bz2. It's twice as big: 17MB uncompressed. Is that a problem?
definately, thanks for noticing this! i was in the belief that the conversion was lossless.
fyi, the resulting repo of the cvs2darcs does not pass a darcs check, and cvs2darcs was last modified sometime in 2004. i've contacted the tailor guys and asked them to take a look at the slime repo because there's probably some bug in tailor that makes it skip the first n changes.
of course i hold back my horses until the conversion is fixed.
On Sun, 19 Nov 2006 21:19:14 +0100, "Attila Lendvai" attila.lendvai@gmail.com wrote:
what do you and the others think about (1) moving the darcs repo at http://common-lisp.net/cgi-bin/darcsweb/darcsweb.cgi?r=cl-wdim-slime;a=summa... under /project/slime? and (2) advertising it as the primary repository?
I'd very much prefer if the CVS repository would continue to be the main repository. Yeah, I know, I haven't done much for SLIME in the last months, but sometimes I do and I think I belong to the minority of SLIME users/hackers using Windoze.
Up until now I haven't managed to get darcs working on Windows, with or without Cygwin. Maybe I'm just too dumb for that kind of stuff, but I have better things to do than to install the version control systems du jour each week.
Does usage of plain old CVS hold back SLIME development in any way?
Cheers, Edi.
Tuesday, November 21, 2006, 10:12:44 AM, Edi Weitz wrote:
EW> I'd very much prefer if the CVS repository would continue to be the EW> main repository. Yeah, I know, I haven't done much for SLIME in the EW> last months, but sometimes I do and I think I belong to the minority EW> of SLIME users/hackers using Windoze.
EW> Up until now I haven't managed to get darcs working on Windows, with EW> or without Cygwin. Maybe I'm just too dumb for that kind of stuff, EW> but I have better things to do than to install the version control EW> systems du jour each week.
Just download it and add its directory to PATH. I had no problems installing darcs on Windows. It just works out of the box.
On Tue, 21 Nov 2006 11:55:39 +0300, Timofei Shatrov grue@mail.ru wrote:
It just works out of the box.
Obviously not for me. Otherwise I wouldn't have written what I wrote...
Edi Weitz edi@agharta.de writes:
I'd very much prefer if the CVS repository would continue to be the main repository. Yeah, I know, I haven't done much for SLIME in the last months, but sometimes I do and I think I belong to the minority of SLIME users/hackers using Windoze.
This is a sort of kind of almost me-too. Darcs is working nicely for me on Windows, but I've had my share of problems with it on ppc/darwin.
Does usage of plain old CVS hold back SLIME development in any way?
Hear, hear.
While I'm not dead set against moving to Darcs, can someone please tell me how it will make Slime hacking better?
A real workflow example or two would be good.
Since Slime would still (I assume) work on a central repository model, I don't see a huge benefit -- but maybe that just shows my relative inexperience with Darcs.
Cheers,
-- Nikodemus Schemer: "Buddha is small, clean, and serious." Lispnik: "Buddha is big, has hairy armpits, and laughs."
While I'm not dead set against moving to Darcs, can someone please tell me how it will make Slime hacking better?
A real workflow example or two would be good.
some quick toughts results in these:
no need for handshaping the ChangeLog
no need for cp -r slime slime.foo.bar to separate unrelated changes if you are experimenting with something that needs further testing and is not yet meant to be committed. also no need to move patches between those slime.foo.bar dirs to have a local merged HEAD for daily use and for testing all the pending changes.
darcs rollback (plus darcs move and easy branching, but they are not that important)
no more hand-editing and mailing of cvs diffs (you can even have a slime_public branch that can accept patches through http)
Not that it matters much but I vote for darcs.
While I'm not dead set against moving to Darcs, can someone please tell me how it will make Slime hacking better?
A real workflow example or two would be good.
When I was locally hacking on slime I had some trouble maintaing multiple patches not committed using cvs. And it was really a pain to share patches with my team-mates while keeping my repo up-to-date with the central repository.
Cheers, levy
* Edi Weitz [2006-11-21 08:12+0100] writes:
Does usage of plain old CVS hold back SLIME development in any way?
The desire to move to some "modern" versioning system was mentioned several times on this list. So, CVS might be a problem for some people. I guess, for those without commit rights.
* Nikodemus Siivola [2006-11-21 10:50+0100] writes:
While I'm not dead set against moving to Darcs, can someone please tell me how it will make Slime hacking better?
A real workflow example or two would be good.
You can put the repo on your laptop and continue hacking Slime in your favorite bar/park/sauna :-)
On the technical side, Darcs should make operations like diff and annotate faster. Also deleting directories would be supported. But after hearing Edi's and your comment about Windows/PPC issues and the obvious problems with the converters, I have the impression that Darcs hasn't yet reached "product quality".
Since Slime would still (I assume) work on a central repository model, I don't see a huge benefit -- but maybe that just shows my relative inexperience with Darcs.
On the social side, it should be easier to maintain forks. We had some unpleasant experiences with people who were, well, less than happy when we didn't commit their patches. With Darcs we could at least say: "Maintain your private branch until your stuff is more mature; it's easy". I, for one, am unhappy that Slime has so many features which I don't use/like/understand. If maintaining a fork were easy, I'd remove fuzzy-completion, presentations, and every line of code that spills across the 80th column. But it's probably to much to ask for a tool to solve social problems.
Helmut.
mature; it's easy". I, for one, am unhappy that Slime has so many features which I don't use/like/understand. If maintaining a fork were easy, I'd remove fuzzy-completion, presentations, and every line
it must be a social issue, because all i can do is wonder how can people use slime without the fuzzy completion... :) it's hard to estimate, but i think at least every 20th keypress of mine is the tab key while coding. but then i guess i should also talk about the length of names we are using... :)
and i must note that branching with darcs is only easy as long as there are few conflicts. no text based tool can handle multiple edits of the same point in a file. if we had a structured editor with a version control system that is preserving identities... but that's a different story and starts with "if".
maybe chopping slime.el into more files would help a bit? emacs sources tend to follow a one file per module approach, but having slime.el, slime-fuzzy.el, etc could work (unless i'm missing something).
* Attila Lendvai [2006-11-21 14:56+0100] writes:
maybe chopping slime.el into more files would help a bit? emacs sources tend to follow a one file per module approach, but having slime.el, slime-fuzzy.el, etc could work (unless i'm missing something).
Well, splitting files doesn't simplify the code. It just moves the mess out of sight. The pressure to factorize and write simple code is IMO higher if everything is in a single file. Also the presentation stuff cannot be modularized as easily as the fuzzy completion code. My problem is that the code is too complicated, not that the file is to big.
Helmut.
-
Attila Lendvai -
Edi Weitz -
Helmut Eller -
Levente Mészáros -
Nikodemus Siivola -
Timofei Shatrov