github.com/moarvm/moarvm | IRC logs at colabti.org/irclogger/irclogger_logs/moarvm
Set by AlexDaniel on 12 June 2018.
nwc10 good *, #moarvm 07:59
nine: that Monday thing seems to have cured itself. 08:00
nine It's a miracle! 08:02
Let's hope, it doesn't come back
timotimo i have a bad feeling about this 08:06
nine A little more optimism, please 08:10
nwc10 We should look forward to Friday. :-) 08:13
timotimo i have a bad feeling about thurs 08:16
nwc10 www.youtube.com/watch?v=mGgMZpGYiy8 08:19
MasterDuke pernos.co/? news.ycombinator.com/item?id=25042827 09:21
MasterDuke fyi, i emailed the gmp mailing list yesterday asking about the linking problem the switch-to-gmp branch has. i haven't heard back yet, but i also mentioned this channel. so if someone shows up talking about gmp, that's probably why 16:20
nine As a possible solution for both the int cache and parametrics caches I figured that when precompiling, I could assemble a list of explicitly requested serialization contexts (i.e. modules) and only accept those as dependencies. 17:07
That way it wouldn't matter if modules loaded by e.g. zef create parametrics that get fetched from the cache during precompilation (which currently would add a bogus dependency). 17:08
However, as good as this may sound at first, it brings another problem: not all dependencies are loaded explicitly. We kinda assume that the setting and its dependencies will already be there. 17:09
nine Actually, the same is true for modules - they're cached by the repository implementation 17:15
nine Upper layers seem to have a better picture of what's getting loaded. But it looks surprisingly hard to find out, which SCs are actually loaded. 17:53

Powered by Cro and the Raku® Programming Language.

Please report any issues / comments / feature requests as an issue on App::Raku::Log.

Thank you!