greptilian logo

IRC log for #sourcefu, 2015-05-25

http://sourcefu.com

| Channels | #sourcefu index | Today | | Search | Google Search | Plain-Text | plain, newest first | summary

All times shown according to UTC.

Time S Nick Message
00:57 pdurbin https://travis-ci.org/OSSIA/OSSIA/jobs/55010038 from https://github.com/OSSIA/OSSIA/blob/master/.travis.yml seems to be a good example of testing on Mac OS X from Travis
00:57 pdurbin more at http://docs.travis-ci.com/user/workers/os-x-infrastructure/
00:59 pdurbin ah but "We are not currently accepting multi-OS requests" according to http://docs.travis-ci.com/user/multi-os/ . interesting
01:10 sivoais yeah... I would really like OS X testing for PDL. However, it would be useful to run my own CI anyway
01:24 pdurbin sivoais: run your own? what would you use? Travis? Jenkins?
01:36 sivoais pdurbin: probably Jenkins. I hear good things about that
01:37 pdurbin sivoais: remind me to let you know when I'm done with these slides: http://bl.ocks.org/pdurbin/03a76a26bd325af0b17e
02:06 sivoais ok! :-)
02:18 pdurbin :)
03:03 pdurbin sivoais: added some screenshots: http://bl.ocks.org/pdurbin/raw/03a76a26bd325af0b17e
03:06 * sivoais looks
03:07 sivoais cool. I think I will need Jenkins at my lab too
13:53 pdurbin interesting question posed at http://www.se-radio.net/2015/04/episode-224-sven-johann-and-eberhard-wolff-on-technical-debt/
13:54 pdurbin ... would your rather work on a code base with a great artitecture and no tests or a code base with a not so great architecture and lots of tests?
14:02 aditsu great architecture, instant choice!
14:02 aditsu tests schmests
14:04 pdurbin heh. one of the two guys on the podcast agrees with you :)
14:13 pdurbin aditsu: even if you are new to the project? if you didn't write any of the code?
14:13 aditsu sure, I thought that was assumed
14:14 pdurbin just checking :)
14:14 aditsu if really needed, tests can be added later, but changing the architecture is much harder
14:15 pdurbin yeah
14:15 pdurbin I just like tests, I guess.
14:15 aditsu and working with a "not so great architecture" is just demoralizing
14:16 pdurbin I guess you should look at the code before taking a new job. :)
14:21 pdurbin "knowledge island"
14:21 pdurbin like a bus number
14:34 aditsu huh?
14:35 pdurbin if only one person knows about a subsystem in the code
17:55 pdurbin prologic: have you read this already? http://techspot.zzzeek.org/2015/02/15/asynchronous-python-and-databases/
17:55 pdurbin mentioned at http://www.talkpythontome.com/episodes/show/5/sqlalchemy-and-data-access-in-python
20:44 prologic pdurbin, hmm not I haven't
20:44 prologic but "Python is very, very slow compared to your Database"
20:44 prologic that's when I stopped reading :)
21:52 pdurbin prologic: that's the sqlalchemy guy! have some respect :)
23:17 prologic I just hate reading articles like that
23:17 prologic honestly
23:18 prologic it's basically negative up front
23:18 prologic Python is really not that slow
23:18 prologic your CODE might be!
23:19 prologic far too often I think us dynamic zealot programmers fall into the habit and trap of just stitching crap together without even thinking about performance or how it might be used
23:19 prologic take for example Django
23:19 prologic whilst it's great "mega" framework that does a lot of things
23:20 prologic web apps developed with it can be a bit slower than say web apps written with very specific high performance code in something like oh I dunno circuits.web? or asyncio or even plain 'ol wsgi
23:20 prologic we fall into the trap with dynamic languages of adding abstraction on top of abstraction and layers upon layers
23:21 prologic function and object lookup/creation and exception handling are generally expensive operations
23:21 prologic don't believe me? :) I can show you some benchmarks to make my point :)
23:21 prologic and let's face it; ofc your database is going to be faster
23:22 prologic it's written in C/C++ most likely
23:22 prologic and written with very specific high performance code
23:22 prologic perhaps I'll have another read of the article tonight at home :)
23:50 pdurbin prologic: you make some good points
23:50 pdurbin and it sounds like you made it farther than I did :)
23:50 pdurbin of course I *did* just listen to him talk for an hour :)
23:56 prologic yeah well
23:56 prologic whilst I may have respect for the guy
23:56 prologic SQLAlchemy is an abstraction layer that's quite complex with many parts
23:56 prologic it adds a layer of abstraction ove ryour database
23:56 prologic and yes this costs in terms of performance
23:57 prologic albeit using an implementation of Python such as PyPy **will* help speed things up dramatically because it can compile into machine code all of the "hotspots"
23:57 prologic but my points above are stillv ery valid
23:58 prologic I find a lot of programmers that use dynamic languages such as Python, Ruby, PHP, etc are often very lazy (in that order of least lazy to most lazy)
23:58 prologic or s/lazy/sloppy

| Channels | #sourcefu index | Today | | Search | Google Search | Plain-Text | plain, newest first | summary

http://sourcefu.com