CVS leim tit files

CVS leim tit files

[Karl Chen]

The leim/CXTERM-DIC/*.tit files should be marked as binary (cvs admin -kb)
because I can't compile them on Cygwin where cvs translates "text" files'
newlines. Copying the *.tit files from a tarball worked.

-- 
Karl Chen / quarl@quarl.org

Re: CVS leim tit files

[Eli Zaretskii]

> From: Karl Chen <quarl@hkn.eecs.berkeley.edu>
> Date: Fri, 6 Sep 2002 15:45:58 -0700 (PDT)
> 
> The leim/CXTERM-DIC/*.tit files should be marked as binary (cvs admin -kb)

That would lose many useful CVS features which are available only
with text files (such as "cvs annotate", "cvs diff", etc.).

> because I can't compile them on Cygwin where cvs translates "text" files'
> newlines.

I suggest to solve that on the Cygwin level.  The whole text/binary
nuisance is a Windows/Cygwin problem whose solution shouldn't IMHO
affect users of other platforms.

Re: CVS leim tit files

[Stefan Monnier]

> > From: Karl Chen <quarl@hkn.eecs.berkeley.edu>
> > Date: Fri, 6 Sep 2002 15:45:58 -0700 (PDT)
> > 
> > The leim/CXTERM-DIC/*.tit files should be marked as binary (cvs admin -kb)
> 
> That would lose many useful CVS features which are available only
> with text files (such as "cvs annotate", "cvs diff", etc.).

I don't know where you got this idea, but it is wrong.  Try it for
yourself on the loaddefs.el file which has been marked as -kb.
What it does is that it prevents automatic merging (and it also prevents
keyword expansion and line-end conversion).
If those files are manually updated, then I'd agree that -kb is a bad
idea since automatic merging is rather useful.

> > because I can't compile them on Cygwin where cvs translates "text" files'
> > newlines.
> 
> I suggest to solve that on the Cygwin level.  The whole text/binary
> nuisance is a Windows/Cygwin problem whose solution shouldn't IMHO
> affect users of other platforms.

I wouldn't trust the cygwin people to solve this problem, so maybe
we should investigate to see why the eol convention used in those files
matters at all.


	Stefan

Re: CVS leim tit files

[Eli Zaretskii]

On Wed, 18 Sep 2002, Stefan Monnier wrote:

> > That would lose many useful CVS features which are available only
> > with text files (such as "cvs annotate", "cvs diff", etc.).
> 
> I don't know where you got this idea

From the CVS manual.  It only talks about "cvs diff", but I recall that 
RCS had problems with line-oriented info, so I assumed annotate would 
lose as well.  Sorry if I was wrong.

> > I suggest to solve that on the Cygwin level.  The whole text/binary
> > nuisance is a Windows/Cygwin problem whose solution shouldn't IMHO
> > affect users of other platforms.
> 
> I wouldn't trust the cygwin people to solve this problem

I didn't mean that Cygwin maintainers should do that.  I meant that the 
solution should be specific to Cygwin, either in the Emacs code or in the 
way the Cygwin port of the CVS client is set up, or maybe even in the way 
filesystems are ``mounted'' by Cygwin on the user's machine.

> maybe
> we should investigate to see why the eol convention used in those files
> matters at all.

Unless someone already knows why, I fully agree.

Re: CVS leim tit files

[Stefan Monnier]

> > > That would lose many useful CVS features which are available only
> > > with text files (such as "cvs annotate", "cvs diff", etc.).
> > I don't know where you got this idea
> 
> From the CVS manual.  It only talks about "cvs diff", but I recall that 
> RCS had problems with line-oriented info, so I assumed annotate would 
> lose as well.  Sorry if I was wrong.

I assume you mean "with non-line-oriented info".
The problem is not that it doesn't work, but that it still assumes
that lines are a meaningful unit, which is not the case for gzipped files
for example, so the result will be poor (tho correct).
But as long as the file is line-oriented, `cvs diff'
and `cvs annotate' work just fine whether or not you use -kb.

> I didn't mean that Cygwin maintainers should do that.  I meant that the
> solution should be specific to Cygwin, either in the Emacs code or in the
> way the Cygwin port of the CVS client is set up, or maybe even in the way
> filesystems are ``mounted'' by Cygwin on the user's machine.

Sorry, I indeed misunderstood you.


	Stefan