| 1 |
Il giorno gio, 10/03/2011 alle 10.56 +0100, justin ha scritto: |
| 2 |
> |
| 3 |
> |
| 4 |
> just because upstream has this configure flag. but it seems we agree |
| 5 |
> on |
| 6 |
> removing the flag completely, defaulting on largefile support and |
| 7 |
> fixing |
| 8 |
> what needs a fix. |
| 9 |
> Correct? |
| 10 |
|
| 11 |
Correct. Any configure with AC_SYS_LARGEFILE will get a |
| 12 |
--enable-largefile option, but if you grep through the tree, this is |
| 13 |
usually simply added unconditionally. |
| 14 |
|
| 15 |
It is more interesting to see whether it actually supports LFS entirely; |
| 16 |
quite a few packages forget to include config.h in some file or other, |
| 17 |
and cause mixed LFS and non-LFS syscalls to be used, which is bad. I |
| 18 |
have written an utility as part of Ruby-Elf[1] called verify-lfs that is |
| 19 |
designed to check for that. |
| 20 |
|
| 21 |
Also note that largefile has no meaning on most 64-bit arches (AMD64 for |
| 22 |
sure, I guess the others as well). |
| 23 |
|
| 24 |
[1] http://www.flameeyes.eu/projects/ruby-elf |
| 25 |
|
| 26 |
-- |
| 27 |
Diego Elio Pettenò — Flameeyes |
| 28 |
http://blog.flameeyes.eu/ |
Archives