#twisted

new core enhancement https://tm.tl/#9069 by rodrigc: Replace xrange with twisted.python.compat.xrange

rodrigc submitted <https://tm.tl/#9069>; - Replace xrange with twisted.python.compat.range (unassigned) for review

new core enhancement https://tm.tl/#9070 by rodrigc: Add twisted.python.compat.reload

Tickets pending review: https://tm.tl/#8566, #8966 (the0id), #9006, #9028, #9040, #9042, #8748, #9047, #9065, #9066, #9067, #9069

new core enhancement https://tm.tl/#9071 by rodrigc: Remove use of apply() in src/twisted/protocols.dict.py

new core enhancement https://tm.tl/#9072 by rodrigc: pyflakes3 build in tox should check more files

rodrigc submitted <https://tm.tl/#9072>; - pyflakes3 build in tox should check more files (unassigned) for review

rodrigc submitted <https://tm.tl/#9070>; - Add twisted.python.compat.reload (unassigned) for review

rodrigc submitted <https://tm.tl/#9071>; - Remove use of apply() in src/twisted/protocols.dict.py (unassigned) for review

new core enhancement https://tm.tl/#9073 by rodrigc: Fix a few places where unicode is used, to fix pyflakes on Python 3

new core enhancement https://tm.tl/#9074 by rodrigc: Update tox.ini to use twistedchecker 0.7.1

Tickets pending review: https://tm.tl/#8566, #8966 (the0id), #9006, #9028, #9040, #9042, #8748, #9047, #9065, #9066, #9067, #9069, #9072, #9070, #9071

rodrigc submitted <https://tm.tl/#9074>; - Update tox.ini to use twistedchecker 0.7.1 (unassigned) for review

exarkun reviewed <https://tm.tl/#9071>; - Remove use of apply() in src/twisted/protocols.dict.py (assigned to rodrigc)

exarkun reviewed <https://tm.tl/#9069>; - Replace xrange with twisted.python.compat.range (assigned to rodrigc)

exarkun reviewed <https://tm.tl/#9028>; - Probably docstring of flattenString references wrong IO class (assigned to exarkun)

Jean-Paul Calderone <exarkun@users.noreply.github.com>; closed <https://tm.tl/#9028>; - Probably docstring of flattenString references wrong IO class

exarkun reviewed <https://tm.tl/#9065>; - Use less restrictive pin for idna. (unassigned)

exarkun submitted <https://tm.tl/#8368>; - Deprecate twisted.cred.credentials.UsernameHashedPassword (assigned to mesozoic) for review

exarkun reviewed <https://tm.tl/#8368>; - Deprecate twisted.cred.credentials.UsernameHashedPassword (assigned to rodrigc)

Tickets pending review: https://tm.tl/#8566, #8966 (the0id), #9006, #9040, #9042, #8748, #9047, #9066, #9067, #9074, #9072, #9070

Hi guys I tried writing a file download accelerator in twisted ... I create 5 connections with the same peer who has the file and request for unique chunk ranges so as to use the complete link .. however when i ran wireshark in background, it showed many warnings regarding "TCP Zero Window Segment" "Connection Reset" "Transmission Window full" and majority shows "ACKed segment wasn't captured"..

whereas when i use the FileSender module in twisted, wireshark hardly shows any error .. why is that so ?

I cant use FileSender to download chunks of file simultaneously from peers... i wish for a performance like FileSender in my file download accelerator module

could any provide any inputs on how i can achieve good performance in my file sharing application ??

Hi :)

I just need to a run a function that can be interrupted sanely without leak based on a timeout and max_memory. And runs sanely on Linux/macOS/Windows

with a sub processes pool of up to n procs

Is there something in twisted for me?

I mean what have works, but randomly fails in some cases.

I use an outer multiprocessing pool to schedule each function's run and then I run the actual function in a thread. The main loop is a sentinel on timeout and max mem

and the random failure depend on the OS and the python version. :|

note that this is a CLI app. so I cannot ask my users to run more than one thing :)

pombreda: mostly twisted's answer to threading heisenbugs is "don't use threading"

wow, umm, yeah, I doubt that code is salvageable. if you really need to run an arbitrary python function with a timeout, spawning a child process is probably the only reliable option

njs, the bug is this: https://github.com/nexB/scancode-toolkit/issues...

njs, I had earlier version using only subprocesses instead. they were slower, but heck, who cares about speed of code that does not run?

njs, so I guess the only sane thing is to use a Process instead

that's what processes are for, pretty much :-)

reliably isolating and accounting for a piece of computation

I end up having to write such ugly things as this https://github.com/nexB/scancode-toolkit/blob/d...

patching threads, patching pools, .... *sigh* and usually ending up on bugs.python.org where the bugs is only on some version sna dnot other

but some later too.

</rant>

njs: you wisdom is welcomed :) thanks

*your

pombreda: you might find this useful: https://stackoverflow.com/questions/23434842/py...

:)

njs: unrelated since I have your attention: just curious: why not using https://bitbucket.org/ronaldoussoren/macholib (and https://github.com/erocarrera/pefile but I guess you know about it) for https://github.com/njsmith/machomachomangler ?

*blink*

njs, thanks the link is very usefule. You have been there for sure :P

actually let me pm you in the macho question. This is nothing about twisted

exarkun closed <https://tm.tl/#9065>; - Use less restrictive pin for idna.

Tickets pending review: https://tm.tl/#8566, #8966 (the0id), #9006, #9040, #9042, #8748, #9047, #9066, #9067, #9074, #9070, #9072

rodrigc submitted <https://tm.tl/#9069>; - Replace xrange with twisted.python.compat.range (unassigned) for review

rodrigc submitted <https://tm.tl/#8009>; - twisted.web.twcgi should be ported to Python 3 (assigned to Unit03) for review

rodrigc reviewed <https://tm.tl/#9067>; - news entry for the removal of received_headers and has the wrong name in it (assigned to exarkun)