#twisted

Tickets pending review: https://tm.tl/#5406 (MostAwesomeDude), #9305, #9333 (markrwilliams), #9337, #9398, #9399, #9406, #9404, #9408, #9118 (the0id), #4964 (jameshilliard), #9138, #9176

Tickets pending review: https://tm.tl/#5406 (MostAwesomeDude), #9305, #9333 (markrwilliams), #9337, #9398, #9399, #9406, #9404, #9408, #9118 (the0id), #4964 (jameshilliard), #9138, #9176

Tickets pending review: https://tm.tl/#5406 (MostAwesomeDude), #9305, #9333 (markrwilliams), #9337, #9398, #9399, #9406, #9404, #9408, #9118 (the0id), #4964 (jameshilliard), #9138, #9176

just did my first twisted amp example (https://github.com/altendky/altwistendky/blob/d...). i had to return a dict from the responder rather than just a value like the example shows (http://twistedmatrix.com/documents/current/core...). otherwise i get BadLocalReturn (https://gist.github.com/altendky/c2ed59c929a942...). makes sense but seems

that maybe the example is wrong? also, in the command class definition i had to b'success' for the response key name but had to regular 'success' for the key when actually constructing the dict. (py3.6)

runciter: in case you wanted to take a quick glance first

ah ha

altendky: it seems very wrong to return a dictionary with native string keys from a responder

good deal, i renamed my repo so that links broken

here's the link to my actual source that worked. https://github.com/altendky/altwistendpy/blob/d...

`return {'success': True}`

one sec

altendky: oh, right - you've set the type of `text` to Unicode

altendky: err, nope

ok

altendky: http://twistedmatrix.com/trac/ticket/6833

looks like some the blame for this is down to AmpBox being a subclass of dict

and also a public API

amp?

not like google amp right?

jfhbrook: nope, not like that! like this: https://amp-protocol.net/

good, cause google amp is awful

the only thing worse than amp is fbia

apple news is #3

point is, the web was good enough

this reminds me a little of dnode

tried porting this to python once and failed cause I was too junior :) kinda wanna try again but now nobody actually uses dnode so..!

runciter: good fun. so that covers the second issue, yeah? but the first where the example responder returns a single value rather than a dict? `return self.uidCounter` http://twistedmatrix.com/documents/current/core...

that just seems like a bug in the docs

altendky: not sure yet though!

altendky: probably good to file both as tickets

altendky: if you can stomach that :)

runciter: perhaps. i'm a little annoyed right now because i've been digging into several open source submissions recently... i mean really, pipenv can't handle installing from github properly? but anyways, not an appropriate feeling to have even if i do have it.

i keep thinking i'm close with my own code just to hit another road block

altendky: i dunno if i agree that it's not appropriate

altendky: whatever it is, it's a defining part of the open source experience

runciter: they just seem to hit in batches

Tickets pending review: https://tm.tl/#5406 (MostAwesomeDude), #9305, #9333 (markrwilliams), #9337, #9398, #9399, #9406, #9404, #9408, #9118 (the0id), #4964 (jameshilliard), #9138, #9176

runciter: i can go months without needing to fix anything. then i've got six projects to try to fix...

altendky: you just need to work on more or different projects! then it can become daily

altendky: suffice it to say i sympathize :)

i'll file the issues

thanks for finding them!

runciter: that's kind of you. thanks

Hello. I have a question on testing code which calls task.Cooperator.coiterate method. In my test I implemented TestScheduler which uses task.Clock.callLater instead of reactors one.

I submit 3 jobs and advance clock by one _EPSILON. When I run under debugger only one job gets processed. But when I just run my test all three jobs get processed.

in my test all the jobs are synchronous

help_seeker: can you share your code?

let me organize it ...

runciter: Please find a copy of my code at https://gist.github.com/alex8080/d32eca07d21cad...

help_seeker: ok, so a cooperator's execution is controlled by two things: a scheduler and a termination predicate

help_seeker: the scheduler makes sure that the cooperator regularly schedules its tasks, while the termination predicate determines how long *all* its tasks have to run

in other words, suppose you want to iterate over X, Y, and Z; the scheduler makes sure that X, Y, and Z have an opportunity to advance every EPSILON seconds. the termination predicate makes sure that X, Y, and Z are advanced until a certain amount of time has passed

https://github.com/twisted/twisted/blob/trunk/s... specifically

help_seeker: https://gist.github.com/alex8080/d32eca07d21cad...

_TestTerminatorPredicateFactory allows each iterator to be advanced exactly once each the Cooperator schedules its tasks

runciter: Thank you!!!

help_seeker: sure thing - does that make sense?

Yes, perfect explanation!

i'll note that i don't think testing this is very valuable

can you explain why you're testing this?

These are my "learning" tests. Just to make sure I understand how things are working.

help_seeker: that's a good reason for writing them, then :)

<runciter>: I tried to advance clock from inside job call: "job = Mock(side_effect=task.clock)". I expected to achieve the desired result (only one job processed), but it did not help.