|
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 29 September 2017. |
|||
|
00:32
P6Bot joined
01:01
P6Bot joined
01:29
P6Bot joined
01:55
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 | ||
|
01:56
synopsebot joined
02:04
synopsebot joined
03:48
synopsebot joined
03:52
synopsebot joined
04:04
lizmat joined
|
|||
| nine | Oh, I've just learned about dnf whatprovides (on Fedora I presume) which finds out what package you need to provide a missing command, library, or cross-compiler. | 05:43 | |
| apt-file find on Debian | 05:45 | ||
|
06:21
llfourn joined
06:26
domidumont joined
06:30
domidumont joined
09:10
leont joined
|
|||
| timotimo | what's the difference between "whatprovides" and "provides"? | 10:01 | |
|
11:19
synopsebot joined
11:29
leont joined
13:50
llfourn joined
15:03
leont joined
|
|||
| ugexe | one is a question | 15:07 | |
| we could do the same thing, too, if bin/ scripts were declared in meta data | 15:08 | ||
| but from the other direction that also makes it easier for zef to use those avenues as alternative installers | |||
| sjn | o/ | 15:23 | |
| sjn somehow managed to break zef by intalling it | |||
| Failed to open file /home/sjn/.rakudo/install/share/perl6/site/dist/2D978098B8202D9A108354C1DA9A36FDC65AF019: No such file or directory | |||
| seems CompUnit::RepositoryRegistry.run-script is looking for a zef that isn't there | 15:24 | ||
| ugexe | do you update your rakudo, or install new versions from a clean slate? | 15:25 | |
| sjn | I updated rakudo | ||
| isn't is safe to update rakudo now and then? | 15:26 | ||
| ugexe | yeah. i'm trying to think *if* when we made the change to .run-script if this situation could somehow happen | 15:27 | |
| i assume `which zef` points at install/share/perl6/site/bin ? | 15:28 | ||
| sjn | .../install/share/perl6/site/bin/zef, yes | ||
| reinstalling zef from scratch fixed it of course. | 15:29 | ||
| ugexe | i think the wrapper rakudo installs getting changed somehow caused this then | 15:30 | |
| sjn | the wrapper? it's super-simple. it only defers to CompUnit::RepositoryRegistry.run-script | 15:31 | |
| calls it | |||
|
15:32
synopsebot joined
|
|||
| sjn | there's no indication in the wrapper about which compunitrepo is the right one | 15:33 | |
| ugexe | i wonder if its possible that it could be finding a precomp'd zef that points to files that no longer exists while also missing some check on .e | 15:42 | |
| probably somewhere in here github.com/rakudo/rakudo/blob/nom/...on.pm#L374 | 15:45 | ||
| sjn | why does that last for loop call return? | 15:49 | |
| ugexe | because it only returns the newest | ||
| sjn | no need to do a for loop for that, is there? | 15:50 | |
| ugexe | i imagine there are other checks that could also take place there, in which case a for loop starts to make more sense | ||
| but yeah, no need for it as it stands | 15:51 | ||
| i got rid of .scripts to go back to .files in github.com/rakudo/rakudo/pull/1125 fwiw | 15:53 | ||
| sjn | anyhoo, I can imagine that the error I pasted earlier can show up if some index isn't updated/recreated when rakudo is updated/rebuilt | 15:55 | |
| sjn is getting wary about all these rabbit holes | 15:56 | ||
| ugexe | has any group mastered upgrading a language install and having all modules still just-work? | 16:00 | |
| sjn | ugexe: nah, don't worry, I'm just venting a little frustration. I'll keep it to myself now :) | 16:02 | |
| sjn approves about the level of ambition in #perl6-toolchain | |||
| ugexe | i was hoping you knew where we could borrow some inspiration | 16:03 | |
| sjn | nope. no inspiration to find out there, except in how not to do things. :) | 16:04 | |
| mst | ugexe: plenv has a command for "take everything from the old install and reinstall it against the new one" but I'm not sure if that counts | ||
| sjn approves of the ambition to handle detailed (micro-level?) upgrades | 16:06 | ||
| nine | I'd guess that there's a mismatch between some short name lookup file and the installed dists | ||
| Maybe some leftover from an uinstall? | 16:07 | ||
| uninstall even | |||
| Could be that we don't uninstall the lookup files for scripts | |||
| No, we remove all files pointing at the dist, regardless of where they come from | 16:09 | ||
| sjn | ugexe: question: would it be ok with you to change the zef usage statement to have the main keyword (e.g. "install", "upgrade", etc.) to be the second word on each line, instead of burying it somewhere at the end? | 16:11 | |
| ugexe | mst: i've considered that too, but can't help but think the installation order inconsistency would eventually cause problems | 16:12 | |
|
16:12
domidumont joined
|
|||
| ugexe | sjn: sure. i dont even remember what the usage statement looks like anymore :P | 16:14 | |
| sjn | oh, zef's using the generated usage | ||
| ugexe | nah | ||
|
16:15
timo joined
|
|||
| ugexe | well it does if you use a bad command | 16:15 | |
| mst | ugexe: could always log each installation somewhere and replay them in order | ||
| sjn | ugexe: that's the one I'm thinking of, actually | ||
| ugexe | $*USAGE.subst(/foo/, "bar", :g) works, but only inside a MAIN sub (so won't work for bad commands) | 16:18 | |
| nine: what points at a dist (other than the bin wrapper) that isn't hash'd with the compiler version? | 16:22 | ||
| could probably do something like a catch-all sub MAIN(|) { $*USAGE ~~ s/// } | 16:25 | ||
| nine | What would reinstalling accomplish? That's only necessary in Perl 5 because XS modules are compiled and linked against a certain perl version | 17:06 | |
| ugexe: can't think of anything | |||
|
17:30
timo joined
19:53
synopsebot joined
|
|||