Re: [PATCH] prune: heed --expire for stale packs, add a test

Nicolas Pitre [off-list ref] writes:

On Tue, 5 Feb 2008, Johannes Schindelin wrote:

quoted

Follow the same logic as for loose objects when removing stale packs: they
might be in use (for example when fetching, or repacking in a cron job),
so give the user a chance to say (via --expire) what is considered too
young an age to die for stale packs.

Also add a simple test to verify that the stale packs are actually
expired.

Signed-off-by: Johannes Schindelin <redacted>

Acked-by: Nicolas Pitre <redacted>

Nicolas

They are not "stale packs", but temporary files that wanted to
become pack but did not succeed.  Perhaps "stale temporary
packs"?

Shouldn't we do something similar to objects/pack/pack-*.temp
files and objects/??/*.temp that http walker leaves?

Hi,

On Tue, 5 Feb 2008, Junio C Hamano wrote:

Nicolas Pitre [off-list ref] writes:

quoted

On Tue, 5 Feb 2008, Johannes Schindelin wrote:

quoted

Follow the same logic as for loose objects when removing stale packs: they
might be in use (for example when fetching, or repacking in a cron job),
so give the user a chance to say (via --expire) what is considered too
young an age to die for stale packs.

Also add a simple test to verify that the stale packs are actually
expired.

Signed-off-by: Johannes Schindelin <redacted>

Acked-by: Nicolas Pitre <redacted>

Nicolas

They are not "stale packs", but temporary files that wanted to
become pack but did not succeed.  Perhaps "stale temporary
packs"?

Shouldn't we do something similar to objects/pack/pack-*.temp
files and objects/??/*.temp that http walker leaves?

Yep.

Ciao,
Dscho

On Tue, 5 Feb 2008, Junio C Hamano wrote:

They are not "stale packs", but temporary files that wanted to
become pack but did not succeed.  Perhaps "stale temporary
packs"?

Shouldn't we do something similar to objects/pack/pack-*.temp
files and objects/??/*.temp that http walker leaves?

Instead, I think http walker should be made to use the same location and 
filename pattern for its temporary files as the rest of the code.


Nicolas

Nicolas Pitre [off-list ref] writes:

On Tue, 5 Feb 2008, Junio C Hamano wrote:

quoted

They are not "stale packs", but temporary files that wanted to
become pack but did not succeed.  Perhaps "stale temporary
packs"?

Shouldn't we do something similar to objects/pack/pack-*.temp
files and objects/??/*.temp that http walker leaves?

Instead, I think http walker should be made to use the same location and 
filename pattern for its temporary files as the rest of the code.

I concur.  That's much cleaner.