qemu-devel
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [PATCH 6/7] .gitlab-ci.d/windows.yml: Unify the prerequisite package


From: Stefan Weil
Subject: Re: [PATCH 6/7] .gitlab-ci.d/windows.yml: Unify the prerequisite packages
Date: Sat, 29 Oct 2022 18:19:25 +0200
User-agent: Mozilla/5.0 (Macintosh; Intel Mac OS X 10.15; rv:102.0) Gecko/20100101 Thunderbird/102.4.0

Am 29.10.22 um 15:06 schrieb Bin Meng:
Hi Thomas,

On Sat, Sep 24, 2022 at 5:20 PM Bin Meng <bmeng.cn@gmail.com> wrote:

Hi Thomas,

On Sat, Sep 10, 2022 at 8:32 AM Bin Meng <bmeng.cn@gmail.com> wrote:

On Sat, Sep 10, 2022 at 12:32 AM Thomas Huth <thuth@redhat.com> wrote:

On 08/09/2022 15.28, Bin Meng wrote:
From: Bin Meng <bin.meng@windriver.com>

At present the prerequisite packages for 64-bit and 32-bit builds
are slightly different. Let's use the same packages for both.

Not sure whether that's a good idea ... I did that on purpose to save some
few time during compilation (since the Windows jobs are running very long
already) ... did you check whether it makes a difference in the run time now?


Not much difference on the build time. Actually I found after we
switched to single thread build the time did not increase too.

One side note regarding the gitlab shared runner:

It seems the shared runner Windows VM is quite slow. Is it possible to
get a faster VM externally?

Any further comment for this patch?


Ping?

Regards,
Bin


Adding more packages typically is a good idea because it allows compilation of more code features, so the coverage during the build is increased.

But here we have 32 and 64 bit builds for Windows which are less different than on Linux because sizeof(long) is the same for both. Nevertheless patch 5 of the series shows an example where 32 bit builds produced a warning, but 64 bit builds did not. So covering as much code as possible (which requires installation of all required packages) can be useful for build tests. And of course it is also necessary if the generated binaries should support as many features as possible.

On the other hand builds with a reduced number of packages can also be reasonable, not only because they need less resources (build time, energy), but also because they can uncover broken dependencies. And for some applications smaller binaries with less features can be sufficient.

So there is no clear answer for me whether the packages for 32 and 64 bit should be synchronized or not.

I think technically the patch is fine. We can postpone the decision because it is not urgent for the upcoming release.

Stefan

Attachment: OpenPGP_0xE08C21D5677450AD.asc
Description: OpenPGP public key

Attachment: OpenPGP_signature
Description: OpenPGP digital signature


reply via email to

[Prev in Thread] Current Thread [Next in Thread]