[bug #67726] Jobserver specification restricts FIFO implementations

Sam James <[email protected]> Tue, 25 Nov 2025 22:25:30 -0500 (EST)
Newsgroups gmane.comp.gnu.make.bugs
Message-ID <[email protected]>
Follow-up Comment #3, bug #67726 (group make):

[comment #2 comment #2:]
>> For some work we're doing [2][3], we're emulating a FIFO using CUSE/FUSE to
>> allow tracking when consumers disappear to avoid lost tokens.
> 
> 
> This is one of those good intentions that has the potential to cause trouble,
> because this feature lets the other side to not track their tokens. Which
> locks the the other side to your jobserver and a specific file type and
> hinders interoperability.
> 

We're diagnosing it when it happens, though. Otherwise you have no idea who
ate the tokens.

> 
>> After you open the FIFO and consume the token, do not close it until after
>> you've written the token back
> 
> A program that is coded to return the tokens, will not at the same time be
> coded to close the fifo prematurely, right?

No, I think we saw a project that does return it, it just doesn't keep the
fifo open the whole time.


    _______________________________________________________

Reply to this item at:

  <https://savannah.gnu.org/bugs/?67726>

_______________________________________________
Message sent via Savannah
https://savannah.gnu.org/
signature.asc (application/pgp-signature, 228 B)
-----BEGIN PGP SIGNATURE-----

iHUEABYIAB0WIQQk97aszIMMAvLLwm6qLAuaBUf3TgUCaSZzKgAKCRCqLAuaBUf3
TjhpAP9BjbMjbqcJv5STRd8LBoWH76ciurwHdkUkXLZ3xojBhwEAq0ePLd1YJtAU
IyMaWw//obOyGsx1x31lnTC3DByStAs=
=t+9F
-----END PGP SIGNATURE-----