That is, when you create the editor, set password='$2b$12$AJTVpce37LM9Dk93qzRe.eMSw1ivsAmoa037.eS6VXoLAyK9cy0YG' and then when POSTing to the oauth endpoint use password='bb'.
2015-12-20 35451, 2015
stanislas
which one to use when initalizing user ?
2015-12-20 35412, 2015
stanislas
aa, ok
2015-12-20 35420, 2015
stanislas
haven't seen your last message
2015-12-20 35407, 2015
stanislas
should it be unicode ?
2015-12-20 35419, 2015
stanislas
Leftmost: ^
2015-12-20 35449, 2015
stanislas
Leftmost: It works :)
2015-12-20 35453, 2015
stanislas
finally
2015-12-20 35412, 2015
Leftmost
The first one is a hashed and salted password, which is what needs to be stored in the database. (We do this for security reasons, so that someone who gets a hold of the database can't figure out anyone's passwords.) When a user logs in, they send across their password and our code uses the information stored in the database to hash the incoming password and check if it's the right one.
2015-12-20 35420, 2015
stanislas
Leftmost: That's exactly how I have imagined it.
2015-12-20 35404, 2015
opatel99
Leftmost: Does BB have an active loading spinner?
2015-12-20 35415, 2015
opatel99
I would think not
2015-12-20 35445, 2015
stanislas
Leftmost: The deleting line doesn't work still.
2015-12-20 35450, 2015
stanislas
It is reporting 401
2015-12-20 35411, 2015
stanislas
Leftmost: I've commited the changes.
2015-12-20 35457, 2015
LordSputnik
opatel99: it does have a loading spinner, but we don't use it anywhere
2015-12-20 35418, 2015
opatel99
LordSputnik: What about the generating margins...?
2015-12-20 35419, 2015
Leftmost
stanislas, when you POST to the oauth endpoint, it should send a response with an access token. You'll need to use that access token when you hit the DELETE creator endpoint.
2015-12-20 35438, 2015
Leftmost
I don't know exactly how to do that in python, though
2015-12-20 35438, 2015
Leftmost
.
2015-12-20 35436, 2015
LordSputnik
stanislas: yeah, the response for auth will contain an access token, you'll have to include that in the "Authorization" header of the delete request
2015-12-20 35444, 2015
Nyanko-sensei joined the channel
2015-12-20 35416, 2015
LordSputnik
stanislas: (the value of the header should be 'Bearer ' + access_token)
opatel99: the contact links should still be grey, also the search input needs decreasing in size still :)
2015-12-20 35423, 2015
opatel99
LordSputnik: On it
2015-12-20 35433, 2015
stanislas
LordSputnik: I know that there is 'headers' parameter, but it's not specified how I should pass it to delete
2015-12-20 35400, 2015
stanislas
delete(url,headers=[my_header]) >
2015-12-20 35416, 2015
stanislas
delete(url,headers=my_header] ?
2015-12-20 35438, 2015
stanislas
* delete(url,headers=my_header)
2015-12-20 35408, 2015
opatel99
LordSputnik: Search on home page or on the navbar?
2015-12-20 35415, 2015
oliverl has quit
2015-12-20 35415, 2015
stanislas
LordSputnik: My current delete looks like that : response2 = self.client.delete('/creator/{}/'.format(random_instance_gid),headers=Headers([('Bearer', access_token)]))
2015-12-20 35448, 2015
Leftmost
The header's name is Authorization, the content 'Bearer ' + access_token.
2015-12-20 35425, 2015
triggerwarning joined the channel
2015-12-20 35427, 2015
dcentral joined the channel
2015-12-20 35410, 2015
stanislas
Leftmost: like this : response2 = self.client.delete('/creator/{}/'.format(random_instance_gid),headers=Headers([('Authorization', 'Bearer' + access_token)])) ?
2015-12-20 35437, 2015
Leftmost
The space is important, but otherwise I think that'd work.
LordSputnik, Leftmost: I am getting to sleep - it is 2 AM for me now, thanks for help so far :)
2015-12-20 35420, 2015
sle has quit
2015-12-20 35435, 2015
LordSputnik
stanislas: you've found a bug ;)
2015-12-20 35438, 2015
LordSputnik
I think
2015-12-20 35402, 2015
LordSputnik
Ahh OK, that is a bug, but you should cover it with another test (ie. test that the endpoint doesn't error if the user leaves the revision note blank)
2015-12-20 35435, 2015
yeeeargh has quit
2015-12-20 35441, 2015
LordSputnik
For now, you'll want to add some JSON data, with something like '{"revision": {"note": "A Test Note"}'
2015-12-20 35414, 2015
opatel99
LordSputnik: anything else?? Or should I make patch and generate screenshots
2015-12-20 35416, 2015
LordSputnik
opatel99: let's go for a first round of screenshots on the task (zip/tar.gz file please!), then we can more easily see how things look on the pages behind authentication
2015-12-20 35422, 2015
LordSputnik
Don't worry about the code for now
2015-12-20 35456, 2015
LordSputnik
opatel99: I won't be able to check them tonight (1:21AM here) but I'll do my best to go through them in about 10 hours time and comment on the task
2015-12-20 35414, 2015
opatel99
Leftmost: ?
2015-12-20 35430, 2015
Bookzombie has quit
2015-12-20 35432, 2015
opatel99
LordSputnik: Okay. Submitting tonight
2015-12-20 35454, 2015
LordSputnik
Then we can do 1 or 2 more rounds if you're around tomorrow at about 7-10 UTC, and then hopefully finalize things :)
2015-12-20 35435, 2015
opatel99
LordSputnik: Im down with that
2015-12-20 35459, 2015
darwin
games
2015-12-20 35400, 2015
Bookzombie joined the channel
2015-12-20 35428, 2015
CatQuest
"side-scrolling skateboarding video game ".. nah I'll pass, not my tpe of thang. maybe kuno?
2015-12-20 35423, 2015
chirlu`
bitmap: Regarding http://explain.depesz.com/s/tXwa, it reads the whole edit_note table and much of edit, too, essentially everything after reosarevok joined MB.
2015-12-20 35448, 2015
chirlu`
And as “hit=344439 read=467499” indicates, more than half of that data was not in cache.
2015-12-20 35404, 2015
chirlu`
Perhaps we should indeed rename the feature to “Edit notes received in the past three months” or something and have a hard cut-off point in time.
2015-12-20 35449, 2015
chirlu`
Together with an index on post_time, this should make it reasonably fast because only a small part of edit_note needs to be considered.
2015-12-20 35436, 2015
darwin
(agree)
2015-12-20 35409, 2015
chirlu`
Alternatively, if we want all-time edit notes, we could add a materialized view that stores the relationship between edit-note-leaver and original-editor directly.
2015-12-20 35442, 2015
chirlu`
Then add an index on original-editor for that view.
2015-12-20 35447, 2015
chirlu`
With that, we wouldn’t need to join edit_note and edit just to find out who was the original editor for each note.
2015-12-20 35427, 2015
chirlu`
Sort of a multi-table index (that isn’t possible otherwise).
2015-12-20 35440, 2015
opatel99
What is the optimal way to output a patch?
2015-12-20 35403, 2015
chirlu`
Optimal for what goal?
2015-12-20 35450, 2015
opatel99
chirlu`: As in does `git diff > output.diff` suffice for most needs?
2015-12-20 35415, 2015
chirlu`
Really depends on the needs. :)
2015-12-20 35425, 2015
chirlu`
Normally, git format-patch will be preferred.
2015-12-20 35401, 2015
opatel99
chirlu`: Okay will do so
2015-12-20 35439, 2015
LordSputnik
zas: It looks like configuring OAuth2 for discourse won't be a trivial thing
2015-12-20 35402, 2015
LordSputnik
It'll require a couple of days of work by someone who's familiar with Ruby (I can do neither of those!)
2015-12-20 35431, 2015
LordSputnik
It also may require modification of the MB web service, or provision of a new endpoint to serve JSON user data when presented with an access token (if https://github.com/discourse/discourse-oauth2-bas… is used, which seems to be the most promising solution)
2015-12-20 35407, 2015
opatel99
LordSputnik: Leftmost I uploaded all my files. Should I press submit or hold off until we discuss?
2015-12-20 35451, 2015
bitmap
chirlu`: I prefer the materialized table idea
2015-12-20 35408, 2015
chirlu`
Really?
2015-12-20 35447, 2015
opatel99
bitmap: What is a materialized table?
2015-12-20 35415, 2015
bitmap
chirlu`: it's *very* fast on my slave server
2015-12-20 35442, 2015
chirlu`
bitmap: Yes, I’d expect so, but it is also quite complex.
2015-12-20 35419, 2015
bitmap
I mean, it's just a trigger, isn't it?
2015-12-20 35424, 2015
chirlu`
opatel99: Essentially, the result of a SELECT over some tables stored in a new table.
2015-12-20 35444, 2015
chirlu`
bitmap: And a schema change. :)
2015-12-20 35458, 2015
bitmap nods
2015-12-20 35409, 2015
Leftmost
You'd need to update the materialized view every time someone left an edit note, presumably. That could be costly, depending on edit note frequency.
2015-12-20 35424, 2015
Leftmost
(That, or present users with out-of-date information, which could be confusing.)
2015-12-20 35458, 2015
chirlu`
Leftmost: It’s only an INSERT on the materialized view per INSERT on edit_note.
2015-12-20 35431, 2015
chirlu`
Because we don’t allow moving edit notes to a different edit etc.
2015-12-20 35414, 2015
Leftmost
Ahh, didn't realize you could INSERT. I'd been thinking about materialized views for a partial solution to collection edit speed (though I'm not sure how effective it would be).
2015-12-20 35453, 2015
chirlu`
Hm, for collections we would need to store a list of all edits affecting one particular collection.
2015-12-20 35404, 2015
bitmap
chirlu`: not saying we shouldn't restrict the post_time for now though
2015-12-20 35406, 2015
chirlu`
That would really be complicated and also require a lot of storage.
2015-12-20 35424, 2015
bitmap
just that eventually I'd like the page to load in under 10 seconds for people
2015-12-20 35414, 2015
chirlu`
It should do that with a time limit and a post_time index, though.
2015-12-20 35426, 2015
Leftmost
I think I was looking at something other than the edit-collection link, but I don't immediately recall what.
2015-12-20 35419, 2015
bitmap
true, it was closer to 5 or 6 seconds on reosarevok's when I tried it
2015-12-20 35443, 2015
chirlu`
Still quite long, that’s true.
2015-12-20 35444, 2015
chirlu`
Leftmost: Aggregated list of all entities contained in a user’s collections?
2015-12-20 35439, 2015
chirlu`
One thing I’m really looking forward to with the Postgres upgrade (apart from index-only scans) is the slow query log.
2015-12-20 35415, 2015
bitmap
is that log_min_duration_statement?
2015-12-20 35448, 2015
chirlu`
Yes, it exists in 9.1 already, but becomes useable in 9.2.
2015-12-20 35403, 2015
bitmap
ah
2015-12-20 35408, 2015
chirlu`
In 9.1 it only looks at exactly identical queries.
2015-12-20 35439, 2015
bitmap
oh, that does sound useful then
2015-12-20 35448, 2015
bitmap
(the 9.2 one)
2015-12-20 35418, 2015
chirlu`
I.e. you looking up your edits is considered to be a totally different query than me looking up my edits, at the moment.
