Re: [Gnu-arch-users] Case insensitive weirdness.

[Dustin Sallings]

On Jan 11, 2004, at 14:34, David Brown wrote:

> I don't know of any good solutions either, other than trying to  
> convince
> the two largest OS vendors in the world that case insensitive
> filesystems are a bad idea.

	Having just run into a problem related to case sensitivity, I can  
imagine how arch can do its part to at least help.

	I've just had some similar problems where I was working on one of  
those systems with a filesystem that is case insensitive (but case  
preserving).

	On this system, I've created a branch of ViewARCH--tez--0.0.7, but I  
called mine viewarch--py22--0.0.7.  At one point, there was also a  
branch called ViewArch--algernon--0.0.1.

	All of these end up under {arch}/ViewARCH (first match) and split out  
from there, which works well on this filesystem where case doesn't  
matter, but apparently I ended up with something with an archive that  
doesn't make a lot of sense along the way, because when I try to check  
it out on a system with a more sane filesystem (NetBSD), I get the  
following error:

Error calling `vu_lstat' for  
"/usr/home/dustin/prog/,,get.viewarch.1073895908.24142.1/,,next- 
patch.1073895909.24142.5/new-files-archive/./{arch}/viewarch/viewarch-- 
py22/viewarch--py22--0.0.7/address@hidden/patch-log/patch-1" (No  
such file or directory)

	I don't believe it would be unreasonable to have category and branch  
names either be lowercase, or fold to lowercase on the filesystem.   
[insert compatibility argument]

	It seems that rooting the patch trees with the category name may not,  
in general, give the best results.  Both the category and branch names  
(and sometimes even the version numbers) can be somewhat arbitrarily  
chosen by individual developers, so the tree feels a little  
upside-down, not to mention the confusion caused by the three different  
capitalizations of ViewARCH.

	As far as source files, I think that's less of a problem, really, as  
that usually represents the actual stuff developers are working on and  
is more visible to the end user.  The stuff that's not visible, the  
stuff internal to arch, is where a bit of protection can fall.  For  
example, in tagline mode, the capitalization of source files should  
never be an issue.  Can we make it less of an issue when explicit tags  
are used?  Perhaps adding a line to the id file with the name of the  
file that tag represents.  If they don't match, avoid the confusion  
right then.

--
SPY                      My girlfriend asked me which one I like better.
pub  1024/3CAE01D5 1994/11/03 Dustin Sallings <address@hidden>
|    Key fingerprint =  87 02 57 08 02 D0 DA D6  C8 0F 3E 65 51 98 D8 BE
L_______________________ I hope the answer won't upset her. ____________