issues@subversion.tigris.org wrote:
>http://subversion.tigris.org/issues/show_bug.cgi?id=639
>
>
>
>
>
>------- Additional Comments From brane@tigris.org 2003-03-17 16:47 PST -------
>O.K., here's the first patch. Sorry, no log message -- I can't keep my
>eyes open long enough to write one. But here's a quick rundown:
>
>1) Introduce svn_filesize_t (really apr_uint64_t) to all (actually,
>most of) the interfaces that deal with whole file sizes and/or offsets
>into whole files (within the FS, of course). This doesn't touch the
>vdelta and diff code, which is a separate issue.
>
>2) Introduce the concept of a maximum memory object size. Functions
>like svn_fs__rep_contents want to read the whole rep into a string,
>but the size of the rep could be much larger than what can actually be
>allocated. So, instead of dying during allocation, check that the size
>of the rep or string is less than some (presumably sane) limit.
>
>Please review. This compiles without new warnings on Windows, and
>passes ra_local and ra_dav tests. I'd be interested in results on
>other platforms (especially becasue apr_off_t and svn_filesize_t are
>the same on Windows, so I probably didn't catch all mismatches).
>
>
>
--
Brane Čibej <brane_at_xbc.nu> http://www.xbc.nu/brane/
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Tue Mar 18 01:50:00 2003