| [01:56] <olli> hi | |
| [01:57] <rick_h> howdy | |
| [02:21] <nigelb> The private team blogpost is so confusing! | |
| [02:24] <rick_h> heh, all the privacy stuff tends to lean that way I think. It's a hard thing to solve/get right | |
| [02:25] <nigelb> I read twice and it took me 5 minutes to understand what the blog post meant. | |
| [02:25] <nigelb> I'm still not confident yet. | |
| [13:50] <jml> lifeless: a day later than promised: https://code.launchpad.net/~jml/testtools/assert-raises-lambda/+merge/90574 | |
| [13:58] * jml gone | |
| [21:24] <james_w> lifeless, around by any chance? | |
| [22:56] <lifeless> fsvo | |
| [23:01] <james_w> hi lifeless | |
| [23:01] <james_w> I'm a bit confused by oops-tools currently | |
| [23:02] <james_w> I can trigger an oops at http://ec2-107-22-26-52.compute-1.amazonaws.com/pkgme/+oops and I can see it go across amqp and it's apparently inserted in to the db at http://ec2-107-20-69-109.compute-1.amazonaws.com/?oopsid=OOPS-884555d8407b0fe9fc62ce0f34094370 but it says there's no matching oopses | |
| [23:03] <lifeless> select * from oops_oops where oopsid = ' | |
| [23:03] <lifeless> OOPS-884555d8407b0fe9fc62ce0f34094370' | |
| [23:03] <lifeless> ; | |
| [23:03] <lifeless> (e.g. have a direct poke at the DB) | |
| [23:03] <lifeless> james_w: what consumer are you using to pull it off of amqp ? | |
| [23:03] <james_w> amqp2disk | |
| [23:04] <james_w> hmm, maybe it's a setup issue actually | |
| [23:06] <james_w> (1 row) | |
| [23:13] <lifeless> ok, so if a literal lookup matches, in the DB, it is inserted | |
| [23:13] <lifeless> and the view should always be searching on the literal string regardless of any heuristics | |
| [23:18] <james_w> ah, I think it was running amqp2disk with a relative pathname | |
| [23:19] <james_w> the exists() check was failing | |
| [23:21] <james_w> http://ec2-107-20-69-109.compute-1.amazonaws.com/?oopsid=OOPS-c470f2678fd5ff65b48a522ae1a35816 | |
| [23:22] <lifeless> yes | |
| [23:22] <lifeless> nice | |
| [23:23] <lifeless> shiny shiny | |
| [23:23] <james_w> it seems like timeline/oops-tools need to be generalised at bit? | |
| [23:23] <james_w> oops-tools assumes that everything is sql? | |
| [23:24] <lifeless> well | |
| [23:24] <lifeless> the UI certainly claims it does | |
| [23:24] <lifeless> and it does the %s %d substitution on everything | |
| [23:25] <lifeless> OTOH thats working quite well for LP for e.g. librarian downloads and so on - mapping different librarian requests to one pattern for aggregation | |
| [23:25] <lifeless> if you would like to make it better, would definitely love to see that | |
| [23:29] <james_w> I need to work out why the sql hooks aren't working here | |
| [23:30] <james_w> http://ec2-107-20-69-109.compute-1.amazonaws.com/?oopsid=OOPS-304f3e703162fa09f30065c7d582c890 is from when they were | |
| [23:32] <lifeless> it seems like you're finding this a bit tricky to debug / sort out | |
| [23:32] <lifeless> if you have suggestions for making that process easier, +1 | |
| [23:37] <james_w> well, it's complicated by the fact that my internet is crappy today and I'm doing this remotely | |
| [23:37] <james_w> plus timeline_django is pretty damn hacky, so it's going to be fragile | |