FS#6976 - Handling of free space checking..
Attached to Project:
Pacman
Opened by Scott H (stonecrest) - Sunday, 22 April 2007, 15:41 GMT
Last edited by Dan McGee (toofishes) - Saturday, 02 June 2007, 17:18 GMT
Opened by Scott H (stonecrest) - Sunday, 22 April 2007, 15:41 GMT
Last edited by Dan McGee (toofishes) - Saturday, 02 June 2007, 17:18 GMT
|
Details
Related to this thread...
http://bbs.archlinux.org/viewtopic.php?id=32201
Is there anything that can be done about these instances where pacman halts to a stop for minutes while checking for free space? Perhaps set a max time of X seconds on the check for free space and either report an error or assume there's enough free space if the max time is reached? |
This task depends upon
Closed by Dan McGee (toofishes)
Saturday, 02 June 2007, 17:18 GMT
Reason for closing: Deferred
Additional comments about closing: removed free space checking code for the time being.
Saturday, 02 June 2007, 17:18 GMT
Reason for closing: Deferred
Additional comments about closing: removed free space checking code for the time being.
Comment by
Aaron Griffin (phrakture) - Monday,
23 April 2007, 04:42 GMT
It's hard to do a "max time" thing because it's a synchronous
blocking call - that is, unless I was to start a new thread or
something (not a good idea), I see little way to make it
timeout.... I need to think about this.
Comment by Dan McGee (toofishes) -
Tuesday, 24 April 2007, 19:45 GMT
What about something like this?
get_freespace.patch
(0.7 KiB)
Comment by Scott H (stonecrest) -
Wednesday, 25 April 2007, 04:01 GMT
I can't speak to the technical merits of that patch, but for what
it's worth, my share was actually of type "cifs" (mount -t cifs
...). I don't really know how it differs from a samba share, but I
assume it should be excluded too.
Comment by Dan McGee (toofishes) -
Wednesday, 30 May 2007, 12:37 GMT
This was not done very cleanly or correctly the first time around.
It doesn't even check free space by partition. I'm considering
just removing it all together until someone provides a more robust
and stable implementation. Suggestions?
Comment by
Aaron Griffin (phrakture) - Friday,
01 June 2007, 18:40 GMT
I've agreed in my head since the very first time I saw that it
depended on mtab. I mean in all honesty, it's an edge case - sure
it'd be nice, but what the hell? Either way, the front end should
do the checking when it gets a list of packages to install (it
knows the size at that point), so lets kill it in git for now, and
readd a more elegant solution to the frontend.
Comment by Dan McGee (toofishes) -
Saturday, 02 June 2007, 17:17 GMT
Removed in my GIT tree. Should be easy to revert if necessary.