Welcome to the main channel on the development of MoarVM, a virtual machine for NQP and Rakudo (moarvm.org). This channel is being logged for historical purposes.
Set by lizmat on 24 May 2021.
El_Che hi, regarding the build problem lizmat posted, I essentially do "git clone --recurse-submodules github.com/moarvm/moarvm.git && cd moarvm && git checkout 2025.01" before I get the "warning: unable to rmdir '3rdparty/rapidhash': Directory not empty" 17:30
tellable6 2025-01-26T01:39:50Z #raku <[Coke]> el_che 2025.01 releases are on github
El_Che the container is clean, it's a new directory, is the clone && checkout tactic a bad idea? 17:34
lizmat well...maybe: it feels to me that this could well have been the first time that MoarVM had commits *after* the release, but *before* you started building packages 17:37
[Coke] can you try starting the clone in the one tag you care about?
ugexe i doubt it is the first time 17:38
it is just the first time that it involved code that would break the olderish code
[Coke] if it ws a branch you could git clone --branch ... not sure if that works on a tag. 17:39
ugexe the rapid-hash thing is a red herring
the problem is specifically libuv 17:40
[Coke] or maybe don't do the submodule step until you've checked out the tag?
ugexe yeah doing the submodule step after the checkout would probably avoid the issue 17:42
[Coke] (I can duplicate the error, unsurprisingly, with the sample git commands above) 17:43
try: git clone, git checkout 2025.01, git submodule init, git submodule update 17:49
El_Che sorry, I answered in #raku: 18:17
18:47 < El_Che> I'll add --recurse-submodules to the checkout and test
18:53 < El_Che> .that seems to do it, thx for thinking with us for a solution
[Coke]: I missed your answer, but we were thinking amond the same lines :)
ugexe: it was the first time we hit this problem. Coincidence probably, could have happened sooner 18:18
ugexe El_Che: how do you know this is the first time and not just the first time it resulted in a compiler error and thus a visible CI issue? 20:19
I would have expected this same issue to have occured for other e.g. libuv updates immediately after a release 20:20
however for those times the updated libuv (that shouldn't be in whatever release) would have still compiled fine
similarly for other submodules 20:21
timo i guess you can't just "checkout" to before a submodule was introduced, unless you manually rm -rf the submodule itself? or is there a submodule inside rapidhash too?
ugexe El_Che: although to be clear I wouldn't spend any effort going back to actually check them. it is just interesting to ponder 20:26
El_Che ugexe: good point. Let's say this is the first time a build failed because of it. 20:36
[Coke] onward and upward. 20:45
jdv [Coke]: backup? please?1 23:23

Powered by Cro and the Raku® Programming Language.

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

Thank you!