On Dec 23, 2007 6:09 PM, Dan McGee <dpmcgee@gmail.com> wrote:
On Dec 23, 2007 1:08 PM, Xavier <shiningxc@gmail.com> wrote:
Allan McRae wrote:
Hi,
I was looking at FS#7155 (http://bugs.archlinux.org/task/7155) which I believe to be fixed now but during then I notice some left overs from what appears to be a rewrite of the md5sum checking.
In the bug report, it was asking for more information than "error: failed to commit transaction (invalid or corrupted package)". Trying to look at this I noticed that the "invalid of corrupted package" part can only generated by alpm_pkg_checkmd5sum() which is not used anywhere. The "error: failed to commit transaction" part is generated by sync_trans() if alpm_trans_commit() fails which in this case can be followed through to test_md5sum() failing. So it looks like md5sum checking was overhauled at some stage.
So my question is: Is alpm_pkg_checkmd5sum() a leftover of the past and need removing?
I guess so, I didn't find yet when its usage was removed, but I at least found when it was introduced. It was first used in src/pacman/sync.c , with commit b708e448ea3 .
It may not be used by our pacman front end, but note that it is a publicly available library function so could be used by other frontends or users of libalpm. With that in mind, does it seem smart to remove it? A quick glance at what it does seems to indicate it could be useful to verify integrity of packages in the sync cache without having to start a transaction.
I vote for keeping it. It's doing no harm