Re: Proposed CHOST change for the 64bit time_t transition

[Andreas K. Huettel]

Am Donnerstag, 5. September 2024, 20:20:32 CEST schrieb Paul Eggert:
> On 2024-09-05 10:03, Andreas K. Huettel wrote:
> > at least time64 implies largefile, so that will get sorted as side
> > effect.
> 
> Right, but this means the "t" in "t64" is somewhat misleading, as it's 
> not just about time: it also affects off_t, ino_t, etc., effects that 
> are in some cases more important than the time_t effects. It may well be 
> a good idea to use a different suffix, to remind non-experts about these 
> other issues. How about "wsi" for "wider system integers"? E.g., 
> i686-pc-linux-gnuwsi.

... or -gnutf64 (for time and files)?

No objections against such variations in principle, just we should keep it
as simple and straightforward as possible then.

> Is the plan to call these APIs XXXt64 or XXXwsi forever, even after the 
> year 2038? For example, will i686-pc-linux-gnu be withdrawn in a few years?

I'd expect that all the 32bit linux variants will slowly sink into obscurity.
(By 2038, we're probably more concerned with the 64bit to 128bit transition.)

Apart from that, I see no real need for further change later on.

> How would you deal with the problem that i686-pc-linux-gnu would mean 
> one thing in Gentoo and something else in Debian? (This was what I was 
> hinting at when I suggested adding a suffix for 32-bit time_t instead of 
> for 64-bit.)

Well, of course I would hope that I can still convince them that tacking on t64
makes sense. After all, it's easy for them now. ;) Otherwise...

Following the logic of riscv versus riscv64 versus riscv32, one could come up
with a "global" definition of
* t64 suffix = 64bit time and lfs
* no suffix = can be either, your fault for not being informative, distro 
default
* t32 suffix = 32bit time (and ...?)


-- 
Andreas K. Hüttel
dilfridge@gentoo.org
Gentoo Linux developer 
(council, comrel, toolchain, base-system, perl, libreoffice)
https://wiki.gentoo.org/wiki/User:Dilfridge

signature.asc
Description: This is a digitally signed message part.