FS#9299 - checksum validity check on that partially downloaded file
Attached to Project:
Pacman
Opened by Julien MISCHKOWITZ (wain) - Tuesday, 22 January 2008, 11:44 GMT
Last edited by Dan McGee (toofishes) - Monday, 24 March 2008, 04:47 GMT
Opened by Julien MISCHKOWITZ (wain) - Tuesday, 22 January 2008, 11:44 GMT
Last edited by Dan McGee (toofishes) - Monday, 24 March 2008, 04:47 GMT
|
Details
If you ^C the program while it's downloading the source
file, next time you'll try to install the same package
yaourt will fail the checksum validity check on that
partially downloaded file. This is fine, but makepkg should
check if the local and remote file size is mismatched and
propose redownloading the file in such case.
cromo |
This task depends upon
Closed by Dan McGee (toofishes)
Monday, 24 March 2008, 04:47 GMT
Reason for closing: Implemented
Additional comments about closing: Makepkg resumable source downloads to *.part implemented in commit f56f7ff39102dab754573b0bc40dbceb5a8ec301 .
Monday, 24 March 2008, 04:47 GMT
Reason for closing: Implemented
Additional comments about closing: Makepkg resumable source downloads to *.part implemented in commit f56f7ff39102dab754573b0bc40dbceb5a8ec301 .
It is your responsibility to get the md5sums right in a PKGBUILD as a packager, in order to make it easily apparant to later users that something went wrong with their download if the md5sum check fails. As only one person really has to change the checksums (the packager).
Makepkg is not a downloader, it just happens to help you out a bit and use one when it can. This might be acceptable in yaourt but I don't plan on doing anything in makepkg.
Sorry if I sounded grumpy, I just tend to try and play devil's advocate with new features to make sure they are well thought out. Although makepkg is not quite right for this, the 1-to-1 mapping between building/installing with yaourt could make this suggestion worthwhile to implement there.
http://archlinux.org/pipermail/pacman-dev/2008-February/011098.html