<<< Date Index >>>     <<< Thread Index >>>

Re: [patch] Solved problems with file size >2GB



On Monday, 24 April 2006 at 17:43, Bruno Cesar Ribas wrote:
> On Mon, Apr 24, 2006 at 01:08:19PM -0700, Brendan Cully wrote:
> > > > Your patch includes inttypes.h unconditionally, which I could imagine
> > > > not to be portable to all system mutt aims to support.  Mutt could maybe
> > > > make use of Gnulib's inttypes module.  But even that won't guarantee you
> > > > all the stuff that this patch needs.  Anyway this patch needs more work
> > > > (sorry for not having the time to delve any deeper into this).
> > > 
> > > Okay!! I'll take inttypes.h out and make it more portable =) no problem, i
> > > thought that using it would be better, but not all systems have it you are
> > > right.
> > > About more work neeeded, what do  do you mean [without deeper 
> > > information]?
> > > I think i'll make a pattern for off_t types in the code and then i'll send
> > > again =)
> > 
> > Mutt already defines LOFF_T and OFF_T_FMT macros for several
> > operations on large files. Could you not use them for display as well?
> 
> I saw that, but i'll have to redefine it because OFF_T_FMT as "%lld" but
> sometimes i'll need only 'lld' because there are that spaces do align, today
> it is used like : "%%sld",fmt , where fmt is the 'number' but that's easy to
> fix.
> But the main problem was that most variables where only 'long int' or 'int'
> or 'size_t' instead of 'off_t', and some casts to 'long' .
> 
> I had put the 'inttypes.h' because I thought it was a good way because it's a
> system header [and now i found out that not all systems have] and have only
> macros verifying if we passed a 64bit 
> I agree that we could just make those OFF_T macros if we cant find that
> macros header [inttypes.h]

I doubt it would be a big deal to change OFF_T_FMT to not include the
%. As for the other type issues, mutt should already be prepared to
handle them portably. See MUTT_C99_INTTYPES in configure.in (defined
in m4/types.m4).