[dev] Re: [suckless] Migration to git

2012-11-26 Thread Christian Neukirchen
Roberto E. Vargas Caballero k...@shike2.com writes:

 On Mon, Nov 26, 2012 at 11:25:18AM +0100, Christoph Lohmann wrote:
 Greetings comrades,

 I  am proposing a migration of all mercurial to git repositories. Git is
 mature and used by nearly all major OSS  projects.  Mercurial  has  this
 slow  prototyping  dependency  of Python, which is annoying and could be
 removed that way. Of course git is still(?) adding a perl dependency.  I
 hope this can be removed from git.

 Perl dependency in git is present only in no basic commands. You can remove
 it and the core system will follow working.

More specifically, the parts that are perl scripts:

/usr/bin/git-cvsserver
/usr/lib/git-core/git-add--interactive
/usr/lib/git-core/git-archimport
/usr/lib/git-core/git-cvsexportcommit
/usr/lib/git-core/git-cvsimport
/usr/lib/git-core/git-cvsserver
/usr/lib/git-core/git-difftool
/usr/lib/git-core/git-instaweb
/usr/lib/git-core/git-relink
/usr/lib/git-core/git-send-email
/usr/lib/git-core/git-svn

Of these, git-add--interactive is the only one which has reasonable
use not related to legacy systems (and actually one of Git's best
features).  (difftool, instaweb and relink are nice additions, but by no
means essential.  send-email probably can be replaced by using straight
MTA injection.)

I think moving to Git is a good idea.

-- 
Christian Neukirchen  chneukirc...@gmail.com  http://chneukirchen.org




[dev] Re: [suckless] Migration to git

2012-11-26 Thread Christian Neukirchen
pancake panc...@youterm.com writes:

 git is far more confusing than hg.. in fact sometimes i have to check
 my notebook or manpages in order to do something in hg is plain easy.

Do you have some concrete examples?  Maybe it's just a matter of taste,
but I find Git more logical to use than Mercurial.  YMMD.

-- 
Christian Neukirchen  chneukirc...@gmail.com  http://chneukirchen.org




Re: [dev] Re: [suckless] Migration to git

2012-11-26 Thread Pierre Chapuis
 pancake panc...@youterm.com writes:

 git is far more confusing than hg.. in fact sometimes i have to check
 my notebook or manpages in order to do something in hg is plain easy.

 Do you have some concrete examples?  Maybe it's just a matter of taste,
 but I find Git more logical to use than Mercurial.  YMMD.

Git is lower level than hg, it exposes lower level concepts to the user
(for instance the Staging Area). Sometimes that makes it confusing
if you do not understand these concepts.

I think the best way to get to know them is this document:
http://ftp.newartisans.com/pub/git.from.bottom.up.pdf

-- 
Pierre Chapuis




Re: [dev] Re: [suckless] Migration to git

2012-11-26 Thread pancake

On 11/26/12 12:31, Christian Neukirchen wrote:

pancake panc...@youterm.com writes:


git is far more confusing than hg.. in fact sometimes i have to check
my notebook or manpages in order to do something in hg is plain easy.

Do you have some concrete examples?  Maybe it's just a matter of taste,
but I find Git more logical to use than Mercurial.  YMMD.


for example

hg reset is the same as git clean -fXd ; git checkout -f

same for 'hg cat' or hg tags..

hg tags = for a in `git tag`; do git show-ref $a ; done



Re: [dev] Re: [suckless] Migration to git

2012-11-26 Thread Lukas Fleischer
On Mon, Nov 26, 2012 at 03:59:11PM +0100, pancake wrote:
 On 11/26/12 12:31, Christian Neukirchen wrote:
 pancake panc...@youterm.com writes:
 
 git is far more confusing than hg.. in fact sometimes i have to check
 my notebook or manpages in order to do something in hg is plain easy.
 Do you have some concrete examples?  Maybe it's just a matter of taste,
 but I find Git more logical to use than Mercurial.  YMMD.
 
 for example
 
 hg reset is the same as git clean -fXd ; git checkout -f

git-reset(1).

 
 same for 'hg cat' or hg tags..
 
 hg tags = for a in `git tag`; do git show-ref $a ; done

`git show-ref --tags`.



[dev] Re: [suckless] Migration to git

2012-11-26 Thread Christian Neukirchen
Andreas Krennmair a...@synflood.at writes:

 * pancake panc...@youterm.com [2012-11-26 18:00]:
On 11/26/12 17:33, Kurt H Maier wrote:
programming languages, and pretending it's 'simple' because it's broken
sloccount

git = 126.000 C
libgit2 = 37.000
mercurial = 34.000 python + 3000 in C

 if you say that python loc is 2x times the same done in C then we
 should move to libgit2 with a decent frontend that can be just few
 lines of code linking to libgit2 statically.

 37.000 LOCs is still a lot, so.. i would like to see a mercurial
 implementation in C

 If you're arguing on this level, then why not switch to OpenCVS?
 http://www.openbsd.org/cgi-bin/cvsweb/src/usr.bin/cvs/

 Only 18,133 SLOCs of C code.

Why not SCCS?

-- 
Christian Neukirchen  chneukirc...@gmail.com  http://chneukirchen.org




Re: [dev] Re: [suckless] Migration to git

2012-11-26 Thread Andreas Krennmair

* Christian Neukirchen chneukirc...@gmail.com [2012-11-26 20:30]:

Andreas Krennmair a...@synflood.at writes:


* pancake panc...@youterm.com [2012-11-26 18:00]:

On 11/26/12 17:33, Kurt H Maier wrote:

programming languages, and pretending it's 'simple' because it's broken

sloccount

git = 126.000 C
libgit2 = 37.000
mercurial = 34.000 python + 3000 in C

if you say that python loc is 2x times the same done in C then we
should move to libgit2 with a decent frontend that can be just few
lines of code linking to libgit2 statically.

37.000 LOCs is still a lot, so.. i would like to see a mercurial
implementation in C


If you're arguing on this level, then why not switch to OpenCVS?
http://www.openbsd.org/cgi-bin/cvsweb/src/usr.bin/cvs/

Only 18,133 SLOCs of C code.


Why not SCCS?


Jörg Schilling's fork of SCCS has 62,917 SLOCs.




Re: [dev] Re: [suckless] Migration to git

2012-11-26 Thread Kurt H Maier
On Mon, Nov 26, 2012 at 08:46:37PM +0100, Andreas Krennmair wrote:
 * Christian Neukirchen chneukirc...@gmail.com [2012-11-26 20:30]:
 Andreas Krennmair a...@synflood.at writes:
 
  * pancake panc...@youterm.com [2012-11-26 18:00]:
 On 11/26/12 17:33, Kurt H Maier wrote:
 programming languages, and pretending it's 'simple' because it's broken
 sloccount
 
 git = 126.000 C
 libgit2 = 37.000
 mercurial = 34.000 python + 3000 in C
 
  if you say that python loc is 2x times the same done in C then we
  should move to libgit2 with a decent frontend that can be just few
  lines of code linking to libgit2 statically.
 
  37.000 LOCs is still a lot, so.. i would like to see a mercurial
  implementation in C
 
  If you're arguing on this level, then why not switch to OpenCVS?
  http://www.openbsd.org/cgi-bin/cvsweb/src/usr.bin/cvs/
 
  Only 18,133 SLOCs of C code.
 
 Why not SCCS?
 
 Jörg Schilling's fork of SCCS has 62,917 SLOCs.
 
 
That's because he's an idiot.

I know you're all being funny, but I just checked, and sure enough: the
git source tree is bigger than the plan 9 kernel.  It's *more* than an
operating system.