sena_kun ugexe, ping? 12:14
Enabled fetching backends [git path curl] don't understand www.cpan.org/authors/id/J/JN/JNTHN/...8.4.tar.gz 12:42
You may need to configure one of the following backends, or install its underlying software - [wget pswebrequest]
Interesting... 12:43
nine sena_kun: can you run zef --debug install cro? 12:45
sena_kun nine, that's the output with --debug from Blin. I checked the container and _indeed_ there was no wget available... Which is hella odd, but does not seem like a fault of rakudo or zef, so it is probably safe to release. 12:46
This docker image worked just fine for a year of releases or something on different machines and now I just switch to my old garbage laptop and boom, something as essential was not installed into the docker image, creepy. 12:48
afk
nine Ah, docker. That may explain why curl can't download an http URI... 12:53
The whole by-reference vs. by-value situation in NativeCall sucks. We have a workaround for inlined data structures with HAS vs. has (a not quite self documenting syntactical difference). But nothing of this sort exists for passing complex data bytes as arguments or having arrays of them. 13:15
C makes this distinction by having Foo (value) vs Foo* (pointer). Since pointers are very common, we take pointers by default, even when a Type is not qualified as such. sub foo(Foo) translates to void foo(Foo*). 13:17
It would be easy to follow C and have a way to get a pointer to a structure (type). Indeed, one of them is already implemented: Pointer[MyStruct]. Possibilities would be $struct.^pointer or even some new operator so there's less typing involved. 13:21
But when the unqualified name already means "pointer to this thing" there's no obvious way to say that one really means the structure itself. 13:22
MasterDuke NativeCall<ver 2>? 13:24
nine May become unavoidable :/ Even so, support for something like MyStruct[] would take a lot of effort on the VM side. The same assumptions govern the whole stack 13:26
Xliff \o 15:20
.seen dwarring
tellable6 Xliff, I saw dwarring 2021-01-05T18:21:11Z in #raku: * dwarring drinks coffee
Xliff .tell dwarring Does LibXML expose xmlNsPtr to Raku? If so, what compunit do I need to load to get it? Thanks. 15:21
tellable6 Xliff, I'll pass your message to dwarring
lizmat just pushec github.com/rakudo/rakudo/commit/16...2cfc602d53 17:03
Introduce Rakudo::Internals.RERUN-WITH method
lizmat just pushed github.com/rakudo/rakudo/commit/c8...283dbfd859 17:12
Introduce the SIL core development helper class
ugexe sena_kun: you did have curl installed though, and that error message is the result of a pretty basic (pure raku) checkk -- github.com/ugexe/zef/blob/f018e1eb...od#L70-L73 18:30
that one has always felt like raku is occasionally doing something wrong with dynamically loaded modules/classes 18:36
sena_kun I reliably get the error for every possible module when run in Blin docker container that worked for years. At the same time I can install modules on this machine fine. It is not a Rakudo regression, because if I run Blin between 2020.11 and 2020.12, it's the same. 19:18
Anyway, no release today, this whole situation sucks.
ugexe the issue, if its rakudo related, would have existed for years 19:39
ugexe my wild guess is something like: precompilation of dynamically loaded modules (the plugins e.g. curl, wget, etc) occasionally does something wrong in such a way that the code still loads, but the method call fails for some reason related to the wrong precomp bits having happened 19:42
the logic for that check in zef is so straightward otherwise that i dont think the issue is in zef 19:43
if you can continue to reproduce it i'd be interested to see what the value of $uri is inside that fetch-matcher sub above since it apparently doesnt start with 'http' in that situation 19:46
but any other time ive seen this happened the issue would go away after rerunning 19:47
sena_kun I have rebuilt the container number of times, purged mounted data volume. 19:48
Plus a Blin ran at the infra server we have did not go well, taking some very bad time to run. 19:49
sena_kun Anyway, right now I simply have no tech conditions to look at all this and pushing out a release knowing there is something like this is a suicide, so good luck next time. Sorry for pinging you for all this, ugexe++ 19:51
ugexe zef added the fez ecosystem in the last week, so that could potentially slow things down 20:22
lizmat just pushed github.com/rakudo/rakudo/commit/1f...f245fb2f3a 22:41
First test file to test inlining