500? that doesn't seem right to me. the server didn't fail. it should be a 4xx response, since the user did something wrong.
lucifer
it fails if we pass like that string as a uuid to sql query.
reosarevok
Probably should catch it and return 400 bad request then
lucifer
postgres fails the query with invalid uuid and so a 500.
yup indeed, makes sense
mayhem
lucifer: yeah, we should catch that PG exception and turn it into a 400.
lucifer
it might be useful to add type checking to dataset hoster i guess before passing params to postgres.
mayhem
loads for pixelpenguin[m] to do already. :)
I guess that would mean adding types, other than strings.
but yes, agreed, adding types would be good.
lucifer
yeah, i think uuid, numerics and strings would be sufficient for now. maybe boolean too?
mayhem
because then we can trim a UUID field if we find whitespace
lucifer
yeah indeed
mayhem
yes, bool, please.
lucifer
yeah i actually have a few ideas about ds hoster. mostly about how we should separate ui and api parts. i'll try to write it down and ping you later this week.
mayhem
ok, sounds good. that was always the trickiest/messiest part.
lucifer
that would also solve rain0r's woes who was trying to experiment with similar artists/recordings data.
mayhem
yes.
oh, while we're dreaming things up, a dropdown field would be a killer feature.
for selecting the various datasets to query.
lucifer
oh yeah indeed. that's on the lists of things to do.
mayhem
question remaining: how do we populate this field?
lucifer
a bit trickier to model if we want to allow dynamic items in the dropdown.
mayhem
DB discovery or update the hoster def?
heh, great minds.
lucifer
hehe lol
db discovery i'd say.
mayhem
when? startup>
lucifer
i think every query time.
mayhem
because querying for the different versions can be quite slow.
I've done it when I've forgotten all the options.
lucifer
we can cache the query name somewhere else then.
mayhem
then each query would take a few minutes.
lucifer
but i think ds hoster should allow dynamic discovery in general.
mayhem
agreed.
lucifer
upto each individual query to make sure it remains fast enough.
mayhem
a cache entry that is manually invalidated would work fine too
lucifer
yeah indeed.
from ds hoster perspective it should't matter whether we load from redis or PG.
mayhem
agreed.
lucifer
i guess this all will actually make it a 350 hours project
mayhem
possibly. lets get the complete list going and then we'll see.
lucifer
yup sounds good
mayhem
but it would be nice to really turbo charge that project since we've gotten so much use from it.
it still has loads of sharp edges.
lucifer
indeed indeed.
it could also use more tests i think.
mayhem
and the good thing is that with these improvements, we can run troi as a query!
lucifer
ah true indeed!
mayhem
hosted troi. not so sure that is a greatest idea ever, but we'll see.
lucifer
oh on that note, do we want to put migrating daily jams to spark as a project idea?
mayhem
do you think that would make a good project? I have a feeling you would spend 3x your time mentoring someone to do it than it would take you to hammer it out. I'm less enthusiastic about this than the others.
another side note: I've starting playing around with regular expressions for detuning and two regexps alone can do a lot of parsing. I think this is a very promising approach.
oh. let me contact them about that.
lucifer
yeah, actually thinking about it. it would be simpler to do it myself.
oh noice!
mayhem
mail to google sent.
yeah, the regexps are gnarly, but they work really well.
lucifer
thanks
indeed, better than having an entire ai engine for detuning though
mayhem
crafting one takes quite some time, but with the help of regex101.com it is totally workable.
mayhem goes for a walk and will do more work later.
reosarevok: There is probably something to be said about cache change.
(Oops, good walk)
BrainzGit
[listenbrainz-android] 14dependabot[bot] opened pull request #99 (03main…dependabot/gradle/com.github.bumptech.glide-compiler-4.15.1): Bump com.github.bumptech.glide:compiler from 4.15.0 to 4.15.1 https://github.com/metabrainz/listenbrainz-andr...
[listenbrainz-android] 14dependabot[bot] opened pull request #100 (03main…dependabot/gradle/com.github.bumptech.glide-glide-4.15.1): Bump com.github.bumptech.glide:glide from 4.15.0 to 4.15.1 https://github.com/metabrainz/listenbrainz-andr...
reosarevok
yvanzo: I'm not quite sure what to say - I think most people didn't even notice the issue (we never got a ticket from someone else about it at least)
But if you have some idea, maybe you can add it and I'll review it in a bit? :)
yvanzo
Yes, I’m just looking into instructions to clear the cache on mirrors.
reosarevok
I'm not sure the issue affected anything outside editing pages which mirrors don't use, but maybe bitmap can help with instructions once around
adhawkins has quit
adhawkins joined the channel
bitmap
yvanzo: on mirrors entities are only cached for 1 hour by default, so it probably won't be necessary to clear the cache, but you can do it via the redis 'flushall' command (`redis-cli -p $PORT flushall` should work, but if there isn't a separate redis store, it will also clear logins)
I also only noticed the errors on edit pages, not when viewing data or using the web service
I went ahead and tried out adding functionality for adding ISO codes and Links to other entities over the weekend, and looks like Mechanize doesn't even detect the control/text-box required for adding links as shown in the following screenshot:
Fin. And the SEC Award of the week goes to... monkey!
monkey: As an award-winner, what are your impressions about the past week?
monkey
:tad:
🎉 even
Hello everyone !
reosarevok
Still on my list: reosarevok, bitmap, zas, mayhem, lucifer, akshaaatt, atj, aerozol, CatQuest, riksucks, jasje, jivte, kellnerd
Pratha-Fish
congrats monkey :D
monkey
last week I spent entirely too much time on updating our test runner (jest) in LB, and looking at our migration path to replace the React testing library (enzyme) that we use with it.
I also finished replacing our modal components in LB. We should now have all features available on all listenCards across the website
aerozol
:0
monkey
I continued with the LB redesign, which you can get a rough preview of on test.listenbrainz.org
Ah, and along with the modal thing I mentioned, started working on LB-1148 as well