Fork me on GitHub
#clojure-uk
<
2017-03-10
>
Sam H11:03:39

anyone have favourite Clojure libs/code snippets that are examples of nice readable Clojure code?

jjl12:03:39

how about flatland's useful?

thomas13:03:39

@shan have a look at some of Zach Tellman's stuff.

thomas13:03:14

don't you just love it when your tests all work fine locally... but on the build machine they just fail, fail and fail.... 😡

mccraigmccraig13:03:10

yes, that's one of my favourites @thomas

dominicm13:03:58

But not potemkin 😛

thomas14:03:55

slowly slowly getting there..... but turn around time is painfully slow.

thomas14:03:09

nor do I actually understand why I am seeing the difference.

agile_geek14:03:41

@thomas could be a number of things but if your tests are at all interdependent you can't guarantee the order that they will run and if the CI server forks and spins up multiple threads you can get issues if your fixtures do set up or tear down at a level that's not per test.

agile_geek14:03:10

Tests that rely on timing i.e. this must happen within x ms or async tests with alts! that expect an order...all that kind of stuff can be non deterministic

agile_geek14:03:40

If it's all pure fn's with no set up or tear down then I'm stumped

thomas14:03:12

something environmental I think... reading in excel files form classpaths. and classpath are always a pain.

thomas14:03:31

but don't ask why I am reading Excel file from Clojure.

otfrom14:03:17

why are you reading Excel from Clojure?

otfrom14:03:23

(there are a few good libs for that)

mccraigmccraig14:03:29

excel as in an actual excel file @thomas, or the one acceptable (utf-16 tsv) excel export format ?

thomas14:03:45

excel as in an actual excel format

thomas14:03:36

Customer insisted that they supply the data in excel files... customer pays....

mccraigmccraig14:03:52

ah, then it is the tears of the angels causing random bit-flips on your CI server

thomas14:03:33

aah ok... let me talk to the angels for a sec and see if they can stop doing that.

Rachel Westmacott14:03:21

I remember once having java code where the tests passed locally, but it wouldn’t even compile in CI

otfrom14:03:37

thomas mccraigmccraig sometimes dealing w/excel is less pain than having the client create the csv file

mccraigmccraig14:03:01

yeah - i wonder if there is a .net service out there somewhere which will do office fileformat conversations for you ?

minimal14:03:20

a fake automatic service that is actually backed by people

thomas14:03:28

maybe http://zamzar.com could do something like that

thomas14:03:55

(which was started by a friend of mine)

thomas14:03:18

ok, building again... fingers crossed

thomas14:03:18

passed all the tests this time... woohooo

otfrom15:03:48

thx (and if that is an abuse of this forum lmk and I'll take it down)

mccraigmccraig15:03:41

what did you use for your topic-detection work (in the MC case-studies) @otfrom ? spark + LDA ?

otfrom15:03:31

mccraigmccraig mallet (and a prototype in R)

otfrom15:03:54

though munging the docs was done in spark (as was classifying docs)

otfrom15:03:03

the created model was done in mallet

otfrom15:03:21

we're doing some more work on that now and should have a release of some stuff around it soon

jasonbell15:03:27

Oh the joys of Mallet.

mccraigmccraig15:03:57

you are not a mallet fan @jasonbell ?

jasonbell15:03:19

No, no, I do like it.

jasonbell15:03:22

It's very good.

mccraigmccraig15:03:21

but the joy of mallet could do with being a bit more unbridled ?

seancorfield17:03:57

I read that as “mullet”… I need more coffee!

otfrom18:03:02

is your np business at the front and party at the back? If so, your results are probably quite ambiguous.

jonpither20:03:14

The better the wine the better the Clojure :-)

korny21:03:48

@thomas docjure is excellent for reading excel - the underlying Java libraries are quite solid. I'd avoid writing excel if possible - I always prefer to write csv, and let the reader load it to excel themselves. (with the caveat that csv is effectively untyped, which can be fun... I've seen major bugs because a column was semi-random alphanumeric, and in the odd case where every digit was a numeric digit, excel loaded it as a number and stripped leading zeroes...

thomas21:03:52

@korny the problem wasn't really docjure, it was classpath problem.

thomas21:03:58

but fixed for now

mccraigmccraig22:03:54

ooo building excel sheets is something i would like to do, even though it would make me feel unclean