#general
- b
bored-art-40741
03/08/2016, 6:02 PMIt's because of this
design pattern, which I also eliminated in that change because it's confusing in addition to being broken_build_hasher - b
bored-art-40741
03/08/2016, 6:03 PMCopy codedef compute_fingerprint(self, target): """ :API: public """ hasher = self._build_hasher(target) return hasher.hexdigest() - b
bored-art-40741
03/08/2016, 6:03 PMerr - b
bored-art-40741
03/08/2016, 6:03 PMOkay backing up again - w
witty-crayon-22786
03/08/2016, 6:03 PMwe tested this. - b
bored-art-40741
03/08/2016, 6:03 PMIt does do nothing, but it has the bug with structural changes - b
bored-art-40741
03/08/2016, 6:03 PMSo I'm back to agreeing 100% with Stu here - w
witty-crayon-22786
03/08/2016, 6:03 PMyes. your bugfix. - b
bored-art-40741
03/08/2016, 6:04 PMThe bugfix should land regardless, and I should go try to back up my claim that resources are still affecting invalidation somehow - w
witty-crayon-22786
03/08/2016, 6:06 PM+1 - b
bored-art-40741
03/08/2016, 6:14 PMHuh, now that I read this I think
is also just completely brokenTaskIdentityFingerprintStrategy - b
bored-art-40741
03/08/2016, 6:14 PMWell... assuming anything other than Ivy and Zinc opted out of structural changes for targets they don't care about - b
bored-art-40741
03/08/2016, 6:14 PMBut I guess nothing else does, so this is fine... ish - b
bored-art-40741
03/08/2016, 6:18 PM@ancient-stone-50795: Can I get you to ship https://rbcommons.com/s/twitter/r/3185/ given the conversation above? - b
bored-art-40741
03/08/2016, 6:18 PMIt seems like master is currently broken as it is for the purpose of invalidating
targetsresources() - b
bored-art-40741
03/08/2016, 6:19 PMAll this does is make it so it uses the correct mechanism for opting out of fingerprinting - a
ancient-stone-50795
03/08/2016, 6:23 PMPlease describe "master is broken" because I don't know that we agree on what is broken and what works. - a
ancient-stone-50795
03/08/2016, 6:23 PMFast builds that don't work don't count as 'fixed' in my book, which is what I'm afraid might happen in our repo - w
witty-crayon-22786
03/08/2016, 6:23 PMresources do not invalidate compilation in master, and haven't for a few months now. - b
bored-art-40741
03/08/2016, 6:23 PM=Stu - a
ancient-stone-50795
03/08/2016, 6:24 PMok. We've been running recent pants builds, so I don't think we are broken. - a
ancient-stone-50795
03/08/2016, 6:24 PMThe one resource I know of that we do care about we have an awful hack in place to force an invalication. - a
ancient-stone-50795
03/08/2016, 6:24 PMSo I will sign off on it. - b
bored-art-40741
03/08/2016, 6:24 PMYou might have a ticking time-bomb - b
bored-art-40741
03/08/2016, 6:24 PMAh or that - w
witty-crayon-22786
03/08/2016, 6:24 PMbut again, see that blog post. this is about more than resources. - b
bored-art-40741
03/08/2016, 6:25 PMI'm still confused about where
targets are synthesized for the cases you're talking about. There should be like, a concrete line of code where that happensresources() - b
bored-art-40741
03/08/2016, 6:25 PMThis invalidation logic only applies to actual
targets in the graphresources() - w
witty-crayon-22786
03/08/2016, 6:25 PMthey are not synthesized. - w
witty-crayon-22786
03/08/2016, 6:25 PMthey are concrete