Last day to vote/reject on proposed EOL names

[John Arbash Meinel]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Mark Hammond wrote:
> On 2/04/2009 8:11 AM, Talden wrote:
>>> IIUC, native is what 80% people want. May be even 95%.
>>
>> we have files in our project for multiple platforms.  Those files need
>> the correct EOLs regardless of which platform the project is checked
>> out on. So we have needs for
>>
>> eol = native (99.9% of text content)
> 
> Yes, this is closer to what I'd expect - although I was expecting a few
> more 9s on the end ;)
> 
>> eol = crlf (a small number of files)
>> eol = lf (a small number of files)
> 
> I'm genuinely interested to know what practical problem is being solved
> by these extra files.  I can think of a number of 'purity' arguments why
> they might appear desirable, but still having trouble coming up with a
> *real* problem, in practice, caused by not having these modes.  I'd like
> to be enlightened so I can concede my ignorance :)
> 
> Cheers,
> 
> Mark

People checking out a project that has VC7 build projects under Linux,
and accessing them via a Samba share, dual-booting, a VM, rsyncing, etc.
I was personally bit by this a while back with CVS. Not to mention a
fair amount of files that someone checked in from an email/tarball from
someone else. Of course there the big problem was that checking *in*
CRLF under Linux would cause you to check *out* CRCRLF on Windows which
was really unfortunate. We can certainly do better than that...

Though I'll mention we end up with a related issue for someone checking
in a file in whatever format who has a different set of rules than the
person checking the format out. (I did notice that Ian uses a regex to
prevent the CRCRLF issue, for which I was happy.)

John
=:->
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.9 (Cygwin)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iEYEARECAAYFAknUDQ4ACgkQJdeBCYSNAAPZ3ACeNCqSEZGLVjiqNptFyIGKMdTW
gBEAn2tAqqHTzdOP5/5x0NJo+AyDdkNI
=tqj4
-----END PGP SIGNATURE-----