> Date: Sat, 15 Nov 2008 15:44:29 +0100 > From: [EMAIL PROTECTED] > > > It's real and I could send you files. I've been dealing with this in > > pdfTeX lately (i.e. enabling LFS). > > Hehe, no thanks, my laptop won't stop thrashing with that big file. :-) > > So, I guess one solution would be to use -D_FILE_OFFSET_BITS=64, > right ? > > There are two separated issues there: > > * File-size related operations. > > The operations using pdf_off_t values should be fine: on POSIX systems > pdf_off_t will map the native off_t type, that will be 32-bit or > 64-bit wide, depending on the system/compiler/etc. In that case we can > compile with _FILE_OFFSET_BITS if it is defined in the system (we can > check for the availability of _FILE_OFFSET_BITS using autoconf). >
Right, I think we should issue some warning when compiling with a 32-bit offset limit anyways. > * Disk-size related operations. > > I would not rely in pdf_off_t for pdf_fsys_get_free_space. I can think > in two alternatives: > > 1) Make pdf_fsys_get_free_space to return the number of free > blocks instead of the number of free bytes, ala statfs. > > 2) Make pdf_fsys_get_free_space to return a 64bit value in any > case. I'm more inclined to the second, returning a 64bit value. just my 2 cents, -gerel
