Camelia, the Perl 6 bug

IRC log for #bioperl, 2012-07-05

| Channels | #bioperl index | Today | | Search | Google Search | Plain-Text | summary

All times shown according to UTC.

Time Nick Message
01:00 genehack joined #bioperl
02:09 gtuckerkellogg joined #bioperl
08:14 leont joined #bioperl
09:55 leont joined #bioperl
10:43 carandraug joined #bioperl
13:04 hash_table joined #bioperl
13:57 rbuels Ulti: yes, it supports comment lines.  if you want to catch parse exceptions, use a eval {}; if( $@ ){ }  construction.
13:58 rbuels Ulti: if by comment lines you mean the stuff after the identifier on the > lines.
13:58 rbuels Ulti: but if you mean some other kind of comment lines, then no, cause those are very, very nonstandard.
13:59 rbuels actually, very, very is probably too strong, since nobody adheres very well to any standards in bioinformatics
14:51 hash_table joined #bioperl
15:26 Ulti even when they adhere to the standard the data will be poisoned somehow anyway ;)   the proble was the exception in general, getting back undef and a warn would have been better. eval() wasn't all I had to do, as SIG{__DIE__} was being handled and swallowed by the web framework I was using
15:26 Ulti instead I had to add in my own local $SIG{__DIE__} and eval, just to see if I had valid sequence
15:26 Ulti it was a lot of boiler plate for the sake of an error message
15:27 Ulti if I just got a warn() and undef from the constructor that would have been a lot easier to handle and just as informative
15:28 Ulti I'm all for exceptions, but in perl exceptions also come hand in hand with being fatal unless you can actually handle them easily
15:30 * deafferret fondly remembers the day he discovered genbank data that wasn't even consistent with itself (ranges that did not, in fact, translate as stated)
15:37 deafferret "die early, die often" and use Try::Tiny a lot :)
15:37 Ulti I dunno Bioinformatics in general has quite useful and clean data for the most part, its nowhere near as bad as geography
15:40 Ulti you mean fail early, fail doesn't mean die
15:40 Ulti an IO library that gives you undef when it cant read or get you some data is totally acceptable
15:40 Ulti if I think its important enough to die I can   or die   on it
15:41 pyrimidine FASTA comments (lines starting with ';') are not supported, primarily b/c we have never seen them used in the wild very often
15:41 Ulti exceptions are mostly a cludge of horror in Perl, eval completely broke code profiling until quite recently too
15:41 Ulti pyrimidine: a comment line can also just be > after the initial >id comment
15:42 Ulti the > ; is chosen from the first character in the file in FASTA
15:42 pyrimidine you mean the description?
15:42 pyrimidine >ID description
15:42 pyrimidine ; comment
15:42 Ulti ;id comment\nSEQUNCESTRING   <-- valid FASTA
15:42 pyrimidine the first is supported
15:43 Ulti >id comment\n>more comment\nSEQUENCESTRING  <--- valid FASTA
15:43 pyrimidine Ulti: there is no standard specification for FASTA
15:43 Ulti sure there is, what FASTA accepts ;)
15:43 pyrimidine correction: there are thousands of specifications for FASTA
15:44 pyrimidine that's what the FASTA *program* accepts, but even in that documentation there is some commentary about using ';'
15:44 pyrimidine IIRC
15:44 Ulti frankly I dont understand why everything isnt just tab files
15:44 Ulti if its not a binary format it should be a tab delimited table
15:45 pyrimidine ah, but then where would the fun be in that? :)
15:45 Ulti the fun would be I could do some dev between my science instead of the other way around
15:45 Ulti the number of sequence formats truly scares me
15:46 Ulti especially if you include all the new nextgen binary read formats
15:46 Ulti every manufacturer with their own format wtf
15:47 pyrimidine welcome to bioinformatics, where everyone believes their format is the best
15:47 Ulti its just sequences of strings with a very limited alphabet it cant be this hard to come up with an acceptable format, if you want annotation of the sequence have that in another standard file... rendering on screen isn't what data formats are about, especially not rendering to a 60 character wide terminal
15:48 Ulti all binary formats should just be bzip2 of whatever standard text format, job done completely
15:48 pyrimidine simple tab-based files only work efficiently to a certain point. They are terrible at more complex data
15:49 Ulti reasons for bzip2 over gzip, it has data chunks you can index and decompress individually
15:49 Ulti pyrimidine yeah but if you start in on XML then you just get a million and one extensions to whatever the standard format is, though I guess you get rendering to screen for free XSLT
15:52 pyrimidine the problem with trying to solve biological problems with a rigid standard format is that something new always comes along (NGS for example, where each tech has a different metric for deciding quality)
15:52 * Ulti gives up complaining, he is just as much part of the problem... I invented my own tabbed format for using with Hadoop >___<
15:53 pyrimidine Ulti: that's precisely the problem.
15:53 pyrimidine that format works for you, but may not work for others
15:53 pyrimidine there simply isn't a common format that will work for everything (no matter how hard Chado tries :)
15:53 Ulti well with Hadoop they have no choice but some kind of tabbed format unless they really want to write everything in Java
15:54 Ulti also is anyone in here part of the effort for the Perl6 translation of BioPerl?
15:54 pyrimidine p/
15:54 Ulti going to say thanks right now for that
15:55 pyrimidine it needs a LOT of attention, haven't had time to work on that in a while
15:55 Ulti going to be so good for bioinformatics once its a bit more efficient... finally people can make proper parsers so all the formats dont matter
15:55 pyrimidine yes, I agree
15:56 pyrimidine Grammars will be *very* handy indeed
15:56 pyrimidine not to mention roles, concurrency, etc
15:56 Ulti especially since they are objects you can inherit from, so all the variants of a more generic format are easily maintained
15:57 * Ulti checks to see what Perl6 did about exceptions...
15:58 Ulti yes, its sane and done properly
15:58 Ulti I love those guys
15:59 Ulti oh wait I didnt read up to the caveat where lots of people argued to get rid of sig die but they kept it in ;____;
17:07 hash_table joined #bioperl
17:14 sheenams joined #bioperl
17:25 pyrimidine sheenams!
17:25 sheenams hello!
17:26 pyrimidine :)
18:41 bro_grammer joined #bioperl
20:55 hash_table joined #bioperl

| Channels | #bioperl index | Today | | Search | Google Search | Plain-Text | summary