00:01 jartycujck67 left, jartycujck38 joined, jartycujck38 left, jartycujck678 joined 00:02 jartycujck678 left
SmokeMachine apogee_ntv: I've been thinking on trying to write something like this (nothing is implemented yet): www.irccloud.com/pastebin/WM2OHNR2/ 00:11
00:19 Guest90 joined 00:24 Guest90 left
apogee_ntv SmokeMachine: Looks workable 00:30
SmokeMachine I’ll try to play with that idea this week when I have some time… 00:52
apogee_ntv SmokeMachine: I'd maybe leave it til v1, I plan to lock the API at v1 and guarantee no BC breaks. 01:00
It should be fairly stable now
But could change subtly 01:01
v1 will launch with a website, tutorial, FAQs etc. 01:02
01:36 sjn left 01:48 hulk joined 01:50 kylese left 02:15 Sgeo joined, hulk left 02:18 kylese joined 02:22 sjn joined 02:58 kst joined
coleman interesting api 03:25
03:33 vasko4535586 left 03:45 vasko4535586 joined 03:48 Sgeo left 03:56 Sgeo joined 03:59 sibl joined, lichtkind__ joined 04:02 lichtkind_ left 04:11 sibl left 04:13 sibl joined 04:20 sibl left, topnep left 04:24 topnep joined 05:13 havochunter joined 05:22 sibl joined 05:43 dustinm` left 05:58 abraxxa joined 06:01 dustinm` joined 06:07 Aedil left 06:10 [r4] joined 06:25 havochunter left 06:41 Sgeo left 07:06 havochunter joined 07:24 soverysour joined, soverysour left, soverysour joined 07:35 sibl left 07:36 annamalai left 07:46 soverysour left 07:50 annamalai joined 07:51 soverysour joined 07:52 dakkar joined 07:56 Ekho left 07:57 sibl joined 07:58 soverysour left
SmokeMachine apogee_ntv: but is that something you think should be part of the Selkie distro, so I should create an PR with that, or should I create a new distro (Selkie::UI) for that? 08:04
08:08 Ekho joined 08:16 sibl left 08:19 sibl joined 08:29 soverysour joined, soverysour left, soverysour joined 08:32 topnep left 08:34 topnep joined 08:42 soverysour left 08:51 soverysour joined 09:00 librasteve_ left 09:19 sibl left, sibl joined 09:24 sibl left
disbot2 <librasteve> Voldenet: many thanks … I think the Junction wins in terms of code clarity 09:43
09:44 librasteve_ joined, sibl joined 09:47 soverysour left, soverysour joined 09:50 dpk left 10:00 sibl left 10:05 Aedil joined 10:12 soverysour left 10:21 soverysour joined, soverysour left, soverysour joined 10:23 annamalai left 10:38 topnep left 10:39 topnep joined 10:46 soverysour left 11:00 soverysour joined 11:06 Aedil left 11:20 annamalai joined 11:24 sibl joined 11:34 sibl left 11:40 soverysour left 11:54 defaultxr left 11:55 ShimmerFairy left, rnddim joined, rjbs left 11:56 kst`` joined 11:57 kst left, ab5tract left, jcallen left 11:58 ab5tract joined, librasteve_ left 11:59 rjbs joined 12:01 defaultxr joined 12:02 jcallen joined 12:14 rnddim is now known as ShimmerFairy
apogee_ntv SmokeMachine: It's a good question, I think it depends on whether Selkie ends up multi-contributor (my hope) or not. :D 12:26
12:31 Aedil joined 12:43 topnep left 12:44 topnep joined
ugexe [This issue is](github.com/raku-community-modules/...ssues/111) is likely caused by github.com/rakudo/rakudo/commit/dc...fa1cba4108 with the problem being there is a OpenSSL::Stack module declaration that contains a OpenSSL::Stack class declaration, and presumably relying on the old behavior 13:08
lizmat ok, so blin should probably pock that up? 13:09
*pick 13:10
ugexe yeah it certianly will
i have an idea for a fix, but first i have $work
lizmat ++ugexe
13:44 kylese left 13:45 kylese joined
disbot2 <antononcube> I was talking to someone who is familiar with Posit's R-Shiny, which can use Markdown templates to generate interactive interfaces. Any markup language is fine with me. 14:13
14:20 donaldh left
disbot2 <librasteve> apogee_ntv: as an outsider, I suggest to add SmokeMachine PR to your Selkie distribution, then bingo multi author :-) 14:51
14:53 [r4] left
apogee_ntv librasteve: He didn't send a PR yet, just proposition (I think), but yes getting more authors would be good. :D 15:00
ugexe i kind of wonder though if it can't be fixed 15:13
OpenSSL expects OpenSSL::Stack::OpenSSL::Stack to exist as OpenSSL::Stack
presumably other distributions do the same thing
github.com/rakudo/rakudo/issues/5854 15:15
"Note that the old compiler frontend is broken in almost the same way! The only difference is that instead of throwing the Redeclaration exception, it silently replaces the module with the class:"
so i guess we have to decide if we should break things to fix this, or if we need to emulate the broken behavior in rakuast 15:16
by "break things to fix this" i mean we accept the current rakudo PR that fixes 5854 (and is currently merged) but breaks OpenSSL and other things that depend on that broken behavior 15:17
ye olde bug compatibility 15:46
maybe we could emit some type of deprecation/warning for the old behavior as a way of avoiding carrying it into rakuast 15:51
something like "class OpenSSL::Stack replaces enclosing module OpenSSL::Stack in the
outer stash. This is probably not what you want; prefer `unit class OpenSSL::Stack;` or choose different names."
actually im not sure if that is a good idea 15:54
since if the author makes the suggested changes then their code won't work anymore until rakuast is default
16:14 soverysour joined
ugexe huh maybe there is a way to rewrite existing code to work in all scenarios. in this case turning `unit module OpenSSL::Stack; class OpenSSL::Stack {...}` into a single `unit class OpenSSL::Stack` declaration 16:23
16:25 librasteve_ joined 16:27 donaldh joined 16:34 dakkar left 17:05 abraxxa left 17:29 merp left 17:31 merp joined
ugexe yeah i think that would work. we revert my fix to the old backend only, and then we detect (in the old backend) when we silently overwrite the namespace and emit a worry similar to what i suggested above with OpenSSL that tells them to use the full namespace (which will work on the old backend and the new frontend). we would continue to not support the silent rewrite in rakuast 18:10
18:11 abraxxa joined 18:14 annamalai left 18:15 annamalai joined
lizmat blin showed 81 failures due to this commit 18:18
(on #raku-dev)
gist.githubusercontent.com/coke/0a...rtlist.txt
ugexe right, im not too surprised. but they are relying on buggy behavior and could be rewritten correctly 18:19
lizmat for sure... but with 81 modules failing, it feels like this change will need a language level barrrier 18:20
so only 6.e would do the correctv behaviour
whether on the old or the new backend
[Coke] if it was a smaller number we could do PRs to each of the modules. 18:22
81 is a lot.
lizmat yeah, and who knows what it would break that is *not* in the ecosystem
ugexe right, what im suggesting is 1) reverting my old frontend change only 2) having the old frontend detect the problematic scenario and emit a worry/deprecation message telling them to use e.g. unit OpenSSL::Stack instead of module OpenSSL::Stack; class OpenSSL::Stack { };`, and finally that we won't support the silent shadowing in the new frontend (i.e. 6.e)
lizmat fwiw, I would be in favour of reverting both backend for now 18:23
to give us more time to think of a proper solution
ugexe the proper solution is what is implemented now 18:24
github.com/rakudo/rakudo/issues/5854 see nines comments
supporting the broken behavior in the new front end is not a proper solution for example 18:25
lizmat I agree there's an issue... but it also feels like we can only fix this in 6.e, and 6.d and earlier will need to continue to support the old (broken) behaviour 18:26
ugexe in my mind any behavior expectations in rakuast is already gated on 6.e but thats arguable. however the longer we wait to nag the ecosystem to fix their code the longer it will be until the rakuast can reasonable be used. 18:29
apogee_ntv It would be nice to get those modules broken down by author, I can see one of mine on there that'll need updating. 18:31
Oh, 3
Ah, possibly broken because Cro::HTTP is.
ugexe yeah, anything depending on OpenSSL 18:32
[Coke] the author is listed in that shortlist.txt file.
it's just not sorted that way 18:33
if you search for zef:apogee you'll find them
(3)
ugexe what is the shortlist.txt file 18:34
or where rather
[Coke] gist.githubusercontent.com/coke/0a...rtlist.txt 18:36
lizmat ugexe: note that RakuAST defaults to 6.e, but would still need to support 6.d behaviour is so requested
*if
[Coke] just the headlines from the blin run (none of the test output, etc.)
weird. I see Temp::Path's URL in there should be raku.land/cpan:UFOBAT/Temp::Path 18:38
18:39 annamalai left 18:40 annamalai joined
ugexe yeah most of those look like they make network requests, i.e. likely depend on OpenSSL 18:41
lizmat ugexe: if you have an OpenSSL fix, then maybe that should be pushed and another blin run done 18:42
before deciding to revert your fixes
ugexe no branch yet, but gist.github.com/ugexe/3ecaa9afdcd9...793bb89060 is what it should be 18:50
github.com/raku-community-modules/...L/pull/112 19:00
19:01 [r4] joined 19:08 havochunter left
ugexe meh i can probably just add the buggy behavior to rakuast and gate the correct behavior on both frontends in an hour or two after work 19:37
which makes sense to me now after the "note that RakuAST defaults to 6.e, but would still need to support 6.d behaviour is so requested" comment 19:42
20:04 librasteve_ left 20:25 [r4] left 20:59 soverysour left 21:09 Aedil left 21:15 abraxxa left 21:54 tobs left 22:47 tobs joined 22:53 soverysour joined 22:57 Sgeo joined, soverysour left

Powered by Cro and the Raku® Programming Language.

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

Thank you!