|
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 12 September 2016. |
|||
|
06:19
domidumont joined
|
|||
| jdv79 | mst: what did you paste me on aug 31st? it expired when i just tried to look at it. | 11:38 | |
| mst: can we use the real CPAN again now? | 11:40 | ||
| ugexe: and is what what prompted the recently cpan/metacpan change in zef? or just my lack of progress:( | 11:41 | ||
| lizmat | jdv79: I think you will still need to wait a few days | 11:43 | |
| shadow.cat/blog/matt-s-trout/the-ps...xperiment/ # jdv79: in case you haven't seen that | |||
| jdv79 | i've been not paying attention for a few weeks so i imagine i've missed alot | 11:44 | |
| lizmat: thanks! | 11:45 | ||
| lizmat | jdv79: there's also the p6weekly.wordpress.com :-) | ||
| which actually has been weekly for most of this year now :-) | |||
| jdv79 | i know. thanks for that as well:) just been too busy to even remember to peek. but i will now. | 11:46 | |
| mst: finally we're moving forward. nice. | 11:55 | ||
| mst | jdv79: note that if you read back through here ugexe was explaining that they think the PSIXDISTS file naming could be better | 13:09 | |
| but I don't think we got as far as a concensus on which 'better' to try next | 13:10 | ||
| ugexe | jdv79: the cpan ContentStorage is from github.com/ugexe/Perl6-ecosystems, i.e. crawls that PSIXDISTS/Perl6 directory and saves all the META info to a single index (cpan.json). The MetaCPAN stuff still exists (just as --metacpan), except it gets its index from the elastic search server on a per request basis | ||
| mst mostly kept saying "yes, I'm sure it could, that's why we're doing an experiment here, let's keep playing until we find one we all like" | |||
| ugexe | p6c, cpan, and metacpan should all work together just fine though, allowing missing dependencies from one to ecosystem to be pulled from another | 13:11 | |
| speaking of, i've one problem with PSIXDISTS currently. the meta file saved in the distributin tar file modifies the `version` field of the meta data | 13:12 | ||
| this means if a distribution is extracted to some directory you have to already know that it came from cpan - i.e. you have to already know to look for version_from_original_meta | |||
| the .meta files are fine doing this, because when you fetch those you already know where you are getting them from | 13:13 | ||
| mst | that was because PSIXDISTS was forcing dev releases to avoid looking 'official' | 13:19 | |
| in case it accidentally didn't bundle a whole dist or something | 13:20 | ||
| I'm not sure if we should drop that yet | |||
| (also, things in git don't always bump thier versions etc.) | |||
| ugexe | but does it need to be changed in the .tar.gz file itself? cant it just be changed in the .meta file? | ||
| mst | eh? | 13:21 | |
| ugexe | only reason it cant be i can think of is due to the file name itself | ||
| mst | the .meta file *is* the file from the .tar.gz | ||
| ugexe | right, but i dont think the meta file from the .tar.gz should be changed | ||
| mst | ... | ||
| I don't understand | |||
| you want the .meta file to be changed, but you don't want the .meta file to be changed | 13:22 | ||
| huh? | |||
| ugexe | the .meta file is already changed. its apparently because the .tar.gz meta is changed but the how is irrelevant. im saying the .meta that cpan uses to index off of can be changed to whatever it needs, but the original meta file should not have its contents changed | 13:23 | |
| mst | the .meta that cpan uses to index off IS THE FILE FROM INSIDE THE TARBALL | ||
| nine | ugexe: PAUSE extracts the META file from the uploaded archive. There's no way to upload it separately | ||
| mst | it is THE SAME FILE | ||
| ugexe | ok, but im not arguing its not | 13:24 | |
| mst | ... | ||
| you keep saying "can we change the .meta file but not the file in the tarball" | |||
| they are the same file | |||
| there is no possibility to change one but not the other | |||
| the .meta file is just a convenient way of getting at the meta file in the tarball without downloading the whole tarball | |||
| ugexe | so the .meta file that is extracted can't be modified. ok | 13:26 | |
| mst | it's extracted server-side | ||
| ugexe | yes | ||
| mst | so if the .meta file was modified that would indicated an integrity problem in CPAN | ||
| and we'd all have a lot of files to checksum | 13:27 | ||
| ugexe | but by changing the version string you have a different integrity problem | 13:28 | |
| im not saying your problem does not exist. im saying this other problem does exists as well | |||
| mst | no | ||
| changing the version string is not a filesystem corruption error | 13:29 | ||
| ugexe | but the immutable distribution gets mutated. you no longer know if $this copy of some distribution is the same as $that copy | 13:30 | |
| mst | PSIXDISTS is building its own tarballs because all it has as sources is a github repo | ||
| there is no mutation. it's building the immutable distribution like that in the first place | 13:31 | ||
| ugexe | in relation to a single content storage i guess. but the content storages cant reliable be used to compliment each other then | 13:33 | |
| nine | mst: do I understand this correctly that this version change is only a temporary safety measure? | 13:34 | |
| mst | yes | ||
| I'm not sure what the big fucking deal is | |||
| it's actually specifically *because* PSIXDISTS isn't yet supposed to look like a 'real' installation source | |||
| in case there's some sort of disaster like missing files in the tarballs | |||
| that doesn't affect the original reposuitory | |||
| ugexe | there is no big deal, i pointed out the current pain point i'm having and you ran with it | ||
| mst | ugexe: I mean, you're asking to break CPAN to 'fix' something that isn't a bug | ||
| no, I ran with correcting all your misunderstandings | 13:35 | ||
| before PSIXDISTS starts shipping non-dev releases | |||
| it would help if people in here had *some* idea how cpan and the toolchain stuff works | |||
| ugexe | spin it however you want | ||
| mst | and clearly you don't yet | ||
| ... | |||
| trying to fix your ignorance is not 'spin', child. | 13:36 | ||
| lizmat | eh... | ||
| mst | I haven't had enough coffee for this. back in a bit. | 13:37 | |
| jdv79 | mst: ok, i didn't get to backlogging yet. | 14:09 | |
| ugexe: cool | |||
|
14:24
perlpilot joined
|
|||
| stmuk_ | maybe it might be worth adding an explicit warning '"tags" : [ "PSIXDISTS"]' or similar (or support: source PSIXDISTS) to the altered meta? | 15:59 | |
| mst | nice idea, yeah | ||
| or x_psixdists: "This is only a test. Do not adjust your aardvark." | 16:00 | ||
| or something to make it obvious it's a testing run | |||
| stmuk_ | maybe 2 tags "PSIXDISTS" and "EXPERIMENTAL" | 16:01 | |
| mst | ranguard: wherte's the code for this? | 16:03 | |
| jdv79: ? | |||
|
16:26
domidumont joined
|
|||
| jdv79 | github.com/perl6modules | 16:42 | |
| mst: ^ | |||
| mst | ta | ||
|
18:22
FROGGS joined
|
|||
| ranguard | mst: your now an owner of that repo | 18:47 | |
| mst | fuck | 18:48 | |
| ranguard | :) | ||
| mst | stmuk_: so, if you can put your notes about meta warnings as an issue please | ||
| ugexe: you were proposing alternative file naming. please file an issue for that | |||
| stmuk_ | is that repo up to date with the current version munging? | 18:51 | |
| 8 months old? | |||
| mst | it got turned off months ago after the first perl5er tantrums | 18:52 | |
| then nobody could be bothered having the arguments except for me, and I kept forgetting to write the fucking blog post, because I suck | |||
| stmuk_ | seems to be | ||
| mst | so, yes, it's the current version | 18:53 | |
| the hiatus was supposed to be rather briefer than that | |||
| on the upside, most of the attempts to tantrum at me this time around were largely spiked by reply of "ok, go back in time five years and convince Andreas Koenig to exclude perl6 from CPAN; otherwise, please stick to constructive suggestions" | 18:54 | ||
| ranguard | mst: as this is all PSIXDISTS's author - dev vs full shouldn't matter? - as people will know that this author should always be SECONDARY to someone releasing their own version? | 19:00 | |
| mst | ranguard: sticking an _ in the version shouldn't make the version numerically different | 19:01 | |
| so I don't see that it matters | |||
|
19:01
domidumont joined
|
|||
| mst | even version.pm finally respects that | 19:01 | |
| ranguard | true dat :) | ||
| mst | basically, if it causes a problem, that means there's a bug in the 6chain | 19:02 | |
| in which case, good, let's shake 'em out | |||
| ranguard | jdv79: metacpan code base has come a long way | 19:12 | |
| v1.metacpan.org and fastapi.metacpan.org/v1 now powered by the 'master' branch of -web and -api, see you in #metacpan if you want more details :) | |||
|
19:16
TimToady joined
19:17
b2gills joined
19:23
moritz joined,
llfourn_ joined,
stmuk_ joined,
sjn_ joined,
ranguard joined,
lizmat joined,
domidumont joined,
ribasushi joined,
perlpilot joined,
FROGGS joined,
[Coke] joined,
autarch joined,
JimmyZ joined,
sivoais joined,
mst joined,
camelia joined,
Cheery joined,
tony-o joined,
DrForr joined,
ugexe joined,
hoelzro joined,
nine joined,
jdv79 joined
19:33
pnu_ joined
20:17
tbrowder joined
|
|||
| jdv79 | ranguard: thank you. i believe soon i'll update and move forward another inch. | 20:24 | |
|
21:30
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 | ||
|
21:33
TimToady joined
|
|||