[Ontology-editors] [Software-group] obo2obo failing

[Chris Mungall]

Very odd.

Can you send Amina and I the exact obo file and osl script. It seems  
like we're using the the same versions by the cvs revision ID, but  
we're running out of things to try so we may as well be 100% sure.  
Also the stdout/stderr (or a truncated version if it loops).

If it's failing with OE2 too then we have more help tracking this  
down. Amina, can you provide Mike instructions to generate log4j  
tracing commands, these might be useful here.

On Feb 10, 2009, at 4:35 PM, Mike Cherry wrote:

> I changed the OSL script and see exactly the same errors on both  
> Solaris and MacOS.  ??
>
> Parse Done!
> Exception in thread "main" java.lang.StackOverflowError
>
> However, maybe this is a hint.  I tried OBO-Edit 2.000-beta39 and  
> with -Xmx2048M in the vmoptions file there was the same behavior.   
> However with I decreased the memory to 512M it worked?
>
> All works with the file from Feb 5th.
>
> -Mike
>
>
> On Feb 10, 2009, at 3:37 AM, Midori Harris wrote:
>
>> I've fixed the go_write file (will commit later today, probably by  
>> the time you get to email) and updated the documentation. Can't  
>> help with the innards of obo2obo, unfortunately!
>>
>> m
>>
>> On Mon, 9 Feb 2009, Chris Mungall wrote:
>>
>>>
>>> OK, but I think we need to fix the obo file too. Can we do this  
>>> with obo2obo or shall I write a perl script? Should be a 2-liner
>>>
>>> I'm thinking that perhaps the comments issue is not linked to  
>>> obo2obo failing on production, as I can't reproduce Mike's error:
>>>
>>>> /Applications/OBO-Edit1/obo2obo -formatversion OBO_1_0  
>>>> obo_format_1_2/gene_ontology.1_2.obo -writecomments -o /tmp/ 
>>>> gene_ontology.1_0.obo -runscript ~/Eclipse/workspace/OBO-Edit/ 
>>>> docs/examplescripts/strip_disjoint_and_replaced_by_tags.osl
>>>> args[0] = |-formatversion|
>>>> args[1] = |OBO_1_0|
>>>> args[2] = |obo_format_1_2/gene_ontology.1_2.obo|
>>>> args[3] = |-writecomments|
>>>> args[4] = |-o|
>>>> args[5] = |/tmp/gene_ontology.1_0.obo|
>>>> args[6] = |-runscript|
>>>> args[7] = |/Users/cjm/Eclipse/workspace/OBO-Edit/docs/ 
>>>> examplescripts/strip_disjoint_and_replaced_by_tags.osl|
>>>> Allowdangling = false
>>>> Parse Done!
>>>> About to write files..., session object count = 28093
>>>> writePath = null
>>>> savePath = [/tmp/gene_ontology.1_0.obo]
>>>
>>> This is on revision 1.188 of the file
>>>
>>> This is running the release version of obo2obo from OE1.1, using  
>>> the osl script from svn head (the same script did not seem to be  
>>> distributed with OE1). This is on OS X so I have a file
>>> 	more /Applications/OBO-Edit1/obo2obo.vmoptions
>>> 	 -Xmx512M
>>>
>>> I believe this is OS X specific, I think for running obo2obo on  
>>> linux the best thing to do is edit the obo2obo file and set -Xmx
>>>
>>> ...but I have a feeling we did this previously for obo2obo on  
>>> production..?
>>>
>>> Might be an idea to check the actual version used into cvs
>>>
>>> On Feb 9, 2009, at 10:50 AM, Midori Harris wrote:
>>>
>>>> Oh crud -- that page is out of date. The only comments we type in  
>>>> nowadays are the first, i.e. "this term was made obsolete because  
>>>> we didn't like the look of it", kind; the others have been  
>>>> superseded in the OBO format 1.2 files by the 'consider' and  
>>>> 'replaced_by' tags. For the 1.0 format files, comments are  
>>>> created by the obo2obo utility as Mike's cronjob runs it.
>>>> I'm on my way out now, but will work on that doc page tomorrow.
>>>> m
>>>> On Mon, 9 Feb 2009, Tanya Berardini wrote:
>>>>> Is there something I could be doing to help sort this out?  I  
>>>>> did several
>>>>> term obsoletions on both Thursday and Friday, following the  
>>>>> protocol and
>>>>> format outlined in http://geneontology.org/GO.usage.shtml#obs.   
>>>>> I was
>>>>> editing and saving in OE1.101.
>>>>> Cheers,
>>>>> Tanya
>>>>> On Mon, Feb 9, 2009 at 9:19 AM, Chris Mungall  
>>>>> <cjm at berkeleybop.org> wrote:
>>>>>> one possibility is that we have just gone over some kind of  
>>>>>> size threshold.
>>>>>> If so we have to up the memory. This can be done directly in  
>>>>>> the obo2obo
>>>>>> script
>>>>>> however, I have a suspicion this may be some oddity to do with  
>>>>>> the
>>>>>> -writecomments required for 1.2 to 1.0. On friday some  
>>>>>> obsoletions in the
>>>>>> ontology were accidentally made 1.0-style, this may be  
>>>>>> confusing obo2obo.
>>>>>> Let's sort that out first and see if that works
>>>>>> On Feb 9, 2009, at 9:12 AM, Mike Cherry wrote:
>>>>>> Hi,
>>>>>>> the obo2obo command is failing.  It worked last week,  
>>>>>>> specifically it
>>>>>>> worked on version 1.185 of the gene_ontology.1_2.obo file.   
>>>>>>> With version
>>>>>>> 1.188 it doesn't work.
>>>>>>> Here is the command.
>>>>>>> obo2obo -formatversion OBO_1_0 /tmp/gene_ontology.1_2.obo - 
>>>>>>> writecomments
>>>>>>> -o /tmp/gene_ontology.1_0.obo -runscript
>>>>>>> /share/go/bin/oboedit/current/docs/examplescripts/ 
>>>>>>> strip_disjoint_and_replaced_by_tags.osl
>>>>>>> ;
>>>>>>> The error message on failng is just:
>>>>>>> Exception in thread "main" java.lang.StackOverflowError
>>>>>>> It there something wrong with that OBO file?
>>>>>>> -Mike
>>>>>>> _______________________________________________
>>>>>>> Software-group mailing list
>>>>>>> Software-group at geneontology.org
>>>>>>> http://fafner.stanford.edu/mailman/listinfo/software-group
>>>
>> _______________________________________________
>> Ontology-editors mailing list
>> Ontology-editors at geneontology.org
>> http://fafner.stanford.edu/mailman/listinfo/ontology-editors
>
>