| -@gerrit:opendev.org- James E. Blair https://matrix.to/#/@jim:acmegating.com proposed: [zuul/zuul] 890591: Add job failure output detection regexes https://review.opendev.org/c/zuul/zuul/+/890591 | 17:43 | |
| -@gerrit:opendev.org- James E. Blair https://matrix.to/#/@jim:acmegating.com proposed: [zuul/zuul] 890591: Add job failure output detection regexes https://review.opendev.org/c/zuul/zuul/+/890591 | 17:44 | |
| -@gerrit:opendev.org- James E. Blair https://matrix.to/#/@jim:acmegating.com proposed: | 17:46 | |
| - [zuul/zuul] 892059: Switch from fb-re2 to google-re2 https://review.opendev.org/c/zuul/zuul/+/892059 | ||
| - [zuul/zuul] 892060: Add configuration support for negative regex https://review.opendev.org/c/zuul/zuul/+/892060 | ||
| -@gerrit:opendev.org- James E. Blair https://matrix.to/#/@jim:acmegating.com proposed: | 19:42 | |
| - [zuul/zuul] 890591: Add job failure output detection regexes https://review.opendev.org/c/zuul/zuul/+/890591 | ||
| - [zuul/zuul] 892059: Switch from fb-re2 to google-re2 https://review.opendev.org/c/zuul/zuul/+/892059 | ||
| - [zuul/zuul] 892060: Add configuration support for negative regex https://review.opendev.org/c/zuul/zuul/+/892060 | ||
| @clarkb:matrix.org | corvus: for 892060 the long term idea is to use re2 (in some implementation) across the board for all regexes due to performance improvements? Then do negative matches "manually" by checking if a positive match fails? | 21:23 |
|---|---|---|
| -@gerrit:opendev.org- James E. Blair https://matrix.to/#/@jim:acmegating.com proposed: [zuul/zuul] 892243: Register RE2 syntax errors as warnings https://review.opendev.org/c/zuul/zuul/+/892243 | 21:54 | |
| @jim:acmegating.com | Clark: yep. the docs have actually just said "we use re2" for a while now, but obviously we have a lot of legacy negative lookaheads, which is not supported by re2. so the "manual" negation check allows that use case to continue. i think it's super useful to be able to say "not master branch" and similar, so that's really been the main holdup. | 21:56 |
| @clarkb:matrix.org | corvus: talking out loud here: but did you consider maybe doing a precompile step on the regex string to chec kif there is a negative lookahead prefix and allow existing regexes to be backward compatible? | 21:57 |
| @jim:acmegating.com | Clark: this new syntax i'm proposing, which is basically: "branches: {regex: master, negate: true}" i this is pretty easy to reed too.... | 21:57 |
| @clarkb:matrix.org | I can think of some reasons to not do that, but am curious if that was considered at all | 21:57 |
| @jim:acmegating.com | Clark: can you take a look at the commit message 892243 and see if that's what you were asking? | 21:57 |
| @clarkb:matrix.org | basically check for `(^?stuff)` and rewrite to `if not re.match(r'stuff')` | 21:58 |
| @jim:acmegating.com | Clark: oh i see what you're saying... | 21:58 |
| @clarkb:matrix.org | ya I guess having warnings would at least tell people what to do to be future compatible. But I'm wondering if we can make the common existing case backward compatible | 21:59 |
| @jim:acmegating.com | Clark: i feel like that's a bit too much like creating our own regex syntax -- i think the warnings period plus deprecation is a good solution | 21:59 |
| @michael_kelly_anet:matrix.org | I see that 9.0.0 dropped last week *and* it added support for gerrit events via kafka. I'm assuming it's just looking for the standard kafka-events stream using the gerrit plugin? Any special considerations when upgrading to 9.0.0? | 21:59 |
| @clarkb:matrix.org | > <@jim:acmegating.com> Clark: i feel like that's a bit too much like creating our own regex syntax -- i think the warnings period plus deprecation is a good solution | 22:01 |
| ya and to be truly generic would involve a lot of parsing. The proposed system is pretty unambiguous without surprising behavior where as what I describe could have a lot of pitfalls if not done right | ||
| -@gerrit:opendev.org- James E. Blair https://matrix.to/#/@jim:acmegating.com proposed: [zuul/zuul] 892247: Fix issue with early failure and blocks https://review.opendev.org/c/zuul/zuul/+/892247 | 23:35 | |
| -@gerrit:opendev.org- James E. Blair https://matrix.to/#/@jim:acmegating.com proposed: [zuul/zuul] 890591: Add job failure output detection regexes https://review.opendev.org/c/zuul/zuul/+/890591 | 23:40 | |
| -@gerrit:opendev.org- James E. Blair https://matrix.to/#/@jim:acmegating.com proposed: | 23:41 | |
| - [zuul/zuul] 892247: Fix issue with early failure and blocks https://review.opendev.org/c/zuul/zuul/+/892247 | ||
| - [zuul/zuul] 890591: Add job failure output detection regexes https://review.opendev.org/c/zuul/zuul/+/890591 | ||
| -@gerrit:opendev.org- James E. Blair https://matrix.to/#/@jim:acmegating.com proposed: | 23:42 | |
| - [zuul/zuul] 892059: Switch from fb-re2 to google-re2 https://review.opendev.org/c/zuul/zuul/+/892059 | ||
| - [zuul/zuul] 892060: Add configuration support for negative regex https://review.opendev.org/c/zuul/zuul/+/892060 | ||
| @jim:acmegating.com | tristanC swest Clark okay i think that series is ready, but we discovered a serious bug, so i extracted the fix into 892247. if you have a sec for an expedited review of that, i'd appreciate it. | 23:43 |
| -@gerrit:opendev.org- James E. Blair https://matrix.to/#/@jim:acmegating.com proposed: [zuul/zuul] 892243: Register RE2 syntax errors as warnings https://review.opendev.org/c/zuul/zuul/+/892243 | 23:44 | |
| @clarkb:matrix.org | corvus: I made a note about a spot that isn't super clear to me | 23:51 |
| @clarkb:matrix.org | but the change looks fine overall | 23:51 |
| @jim:acmegating.com | Clark: responded (and yep i think you got it) | 23:58 |
| @clarkb:matrix.org | thanks your response makes sense to me | 23:59 |
Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!