|
Fire is step THREE! | github.com/perl6/toolchain-bikeshed | Channel logs: irclog.perlgeek.de/perl6-toolchain/today | useful prior art: metacpan.org/pod/CPAN::Meta::Spec Set by moderator on 11 March 2017. |
|||
|
02:48
ilbot3 joined
|
|||
| moderator | Fire is step THREE! | github.com/perl6/toolchain-bikeshed | Channel logs: irclog.perlgeek.de/perl6-toolchain/today | useful prior art: metacpan.org/pod/CPAN::Meta::Spec | ||
|
03:37
ribasushi joined
07:17
perlpilot joined
07:52
domidumont joined
07:57
domidumont joined
|
|||
| nine | ugexe: I wonder if we couldn't just release the lock right after checking if the dist is already installed. | 08:42 | |
| ugexe: CURI is already designed so installation of a dist is only adding new files and not modifying any existing ones. | 08:43 | ||
|
10:53
domidumont joined
|
|||
| ugexe | is that true for the precompilation portion as well though? | 15:41 | |
| like I thought somewhere it just sort of iterated/recursed over everything it needed to precomp until it worked (sometimes doing the same module more than once if its precomp id changes) | 15:42 | ||
| nine | Precompilation must be concurency safe on its own anyway as it can be triggered by simply running programs in parallel (like with prove -j8) | 15:47 | |
| ugexe | i see. but its safe because of the file locking right? I'm not sure where the file lock for precomp ends up, but if its at the root of the precomp repo its not *really* doing anything concurrent other than waiting right? | 15:49 | |
| e.g. -j8 gives you 1 thread with the precomp file lock, and 7 other threads waiting for it (but safely) | 15:50 | ||
| nine | Yes, it right now locks the whole store. Though it should be simple to use more fine grained locking there, i.e. one lock per precompilation file. | 17:26 | |