you use remainder instead of selecting a character of the number as a string
2017-12-22 35600, 2017
alastairp
so 1-999 go in the directory 0000
2017-12-22 35625, 2017
alastairp
wait, I don't think that's exactly right
2017-12-22 35629, 2017
alastairp
no, ignore me
2017-12-22 35638, 2017
ruaok
naiveai: well done!
2017-12-22 35650, 2017
ruaok hopes alastairp will continue
2017-12-22 35656, 2017
alastairp
ruaok: you 0-pad your remainders
2017-12-22 35615, 2017
alastairp
so, %1000 gives you remainders from 0 to 999
2017-12-22 35634, 2017
ruaok
so, the last 4 digits of the filename, sans extension, basically?
2017-12-22 35634, 2017
alastairp
which are your top level directories, padded to 3 or 4 characters
2017-12-22 35659, 2017
alastairp
right
2017-12-22 35601, 2017
github joined the channel
2017-12-22 35601, 2017
github
[sir] samj1912 opened pull request #72: SOLR-77: Fix the way FK rels were handled (master...fixfk) https://git.io/vbHeh
2017-12-22 35601, 2017
github has left the channel
2017-12-22 35630, 2017
samj1912
and that was the last bug from current batch of testing \o/
2017-12-22 35629, 2017
alastairp
I hope we remember that exposing user row-ids leaks internal structure. However, since the data is public I don't think that we have any possible attack vectors
2017-12-22 35602, 2017
ruaok
famous last words, samj1912
2017-12-22 35619, 2017
samj1912
lol :P
2017-12-22 35635, 2017
alastairp
yeah, really sounds to me like editors should have their own uuids
2017-12-22 35636, 2017
ruaok
alastairp: in that case we can use the Auth token UUIDs. We've already got them.
2017-12-22 35640, 2017
ruaok snickers
2017-12-22 35653, 2017
alastairp
ruaok: sounds like you dropped the ball 15 years ago
2017-12-22 35658, 2017
alastairp
didn't think that one through
2017-12-22 35621, 2017
ruaok
my current record on being wrong on something is 20 years.
2017-12-22 35624, 2017
ruaok is proud
2017-12-22 35620, 2017
ruaok
more seriously, the UUID provided from oauth -- can we use that safely or is that a security problem?
2017-12-22 35630, 2017
ruaok
I don't want to generate yet another number.
2017-12-22 35653, 2017
ruaok
then we'd have to store them, otherwise the ids won't be consistent between dumps.
2017-12-22 35609, 2017
alastairp
what do you mean? we make MB give an oauth when we auth and use that?
2017-12-22 35620, 2017
alastairp
that's kind of a solution to the deleted user stuff that we went through
2017-12-22 35627, 2017
ruaok
yep. I think it is already in the DB.
2017-12-22 35638, 2017
alastairp
if it is, then I think that's a pretty nice idea
2017-12-22 35646, 2017
alastairp
not sure if it's exposed though
2017-12-22 35649, 2017
iliekcomputers
MB provides a uuid during oauth?
2017-12-22 35653, 2017
xps2 has quit
2017-12-22 35657, 2017
alastairp
(I'm pretty sure it doesn't)
2017-12-22 35603, 2017
alastairp
but check with bitmap
2017-12-22 35639, 2017
ruaok
yeah, no. we certainly don't store it.
2017-12-22 35649, 2017
ruaok
so, unless we go and give each user a UUID and store it, this won't work.
2017-12-22 35601, 2017
ruaok
we've identified a lot of things that don't work so far. go us.
2017-12-22 35634, 2017
iliekcomputers
anyways, if we were just gonna use uuids, using them for measurement names would have been a good idea too
2017-12-22 35652, 2017
ruaok
saved you some sanity, for sure.
2017-12-22 35600, 2017
alastairp
oh yeah, heh
2017-12-22 35609, 2017
iliekcomputers
:)
2017-12-22 35615, 2017
alastairp
anyway, listenbrainz NGS can use them
2017-12-22 35618, 2017
ruaok
so, what is our conclusion
2017-12-22 35619, 2017
ruaok
?
2017-12-22 35622, 2017
Leftmost
LordSputnik, as it is, it makes no sense to me how the identifier editor populates existing rows.
2017-12-22 35606, 2017
alastairp
use row ids, multiple directories, consider adding uuids when we move oauth to metabrainz.org
