Re: [bug-libunistring] Improve build parallelization patch submitted to vcpkg

[Billy O'Neal (VC AIR)]

>Whether vcpkg applies it or not, is your choice. Whether optimizing the
elapsed time of a build is worth the cost of maintaining a patch that
cannot be upstreamed. Your choice.

That's OK! We just want to ensure that we don't 'speak for upstream' in a way they would find misrepresents them. We'll take the patch.

Thanks for the quick response!

---

From: Bruno Haible <bruno@clisp.org>
Sent: Tuesday, November 15, 2022 07:29 PM
To: bug-libunistring@gnu.org <bug-libunistring@gnu.org>; Billy O'Neal (VC AIR) <bion@microsoft.com>
Subject: Re: [bug-libunistring] Improve build parallelization patch submitted to vcpkg

 

[You don't often get email from bruno@clisp.org. Learn why this is important at https://aka.ms/LearnAboutSenderIdentification ]

Hi,

Billy O'Neal wrote:
> A contributor has asked to apply a patch to libunistring to improve build time by parallelizing some of the operations in the makefile: https://nam06.safelinks.protection.outlook.com/?url="">
>
> Would you be willing to look at this patch and take it or indicate that you don't oppose vcpkg applying it?

As far as I understand the patch [1], it parallelizes a loop that compiles many
source files and collects the symbols defined by the resulting .o files, such
that
  - the total elapsed time will shrink significantly (depending on the number
    of CPUs available),
  - the total CPU time remains the same,
  - there is a code duplication: ca. 12 lines of code are duplicated,
  - the resulting Makefile will not work any more with non-GNU make.

Therefore the patch is not appropriate for upstream.

Whether vcpkg applies it or not, is your choice. Whether optimizing the
elapsed time of a build is worth the cost of maintaining a patch that
cannot be upstreamed. Your choice.

Bruno

[1] https://nam06.safelinks.protection.outlook.com/?url="">