Hi

On Tue, Jul 22, 2025 at 01:00:49AM +0200, Timo Rothenpieler wrote:
> On 7/22/2025 12:17 AM, Michael Niedermayer wrote:
> > On Mon, Jul 21, 2025 at 06:37:06PM +0200, Timo Rothenpieler wrote:
[...]
> > --enable-version3
> > --cc='ccache gcc' (or clang)
> 
> I don't fully trust ccache to not cause spurious issues.

locally i think ive never seen an issue and i use it 100% of teh time
for all builds.

in fate clients, i have seen cases where ccache needed to be cleared
I dont want to guess the cause as its just a guess, but clients
run out of disk space, they run out of memory which doesnt happen
to me locally nearly that often


> Gentoo stopped accepting bug reports if ccache was involved and it wasn't
> reproduced without it.

> The builds are speedy enough that a full build each time doesn't seem too
> horrible.

Being able to tell the developer code is ok/not ok quicker can be valuable
but sure, if you prefer having it disabled for now

> 
> I'd rather just add more runners if we ever run into capacity problems,
> which I honestly don't see happening anytime soon.
> 
> > --assert-level=2
> > --tempprefix=somebasepaththatcanbeusedforcreatingtemporaryfiles
> 
> If I understand this right, all it does is use a fixed prefix in /tmp
> instead of just calling mktemp?
> I don't immediately see the benefit of that, speed wise.

its for ccache. Because without that you have random temporary files
and so they are not cached.

my configure here takes 6sec (with --tempprefix and ccache and lots of 
--enables)
without tempprefix its 10sec

thx

[...]
-- 
Michael     GnuPG fingerprint: 9FF2128B147EF6730BADF133611EC787040B0FAB

While the State exists there can be no freedom; when there is freedom there
will be no State. -- Vladimir Lenin

Attachment: signature.asc
Description: PGP signature

_______________________________________________
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
https://ffmpeg.org/mailman/listinfo/ffmpeg-devel

To unsubscribe, visit link above, or email
ffmpeg-devel-requ...@ffmpeg.org with subject "unsubscribe".

Reply via email to