[Bioperl-l] SearchIO speed up

Chris Fields cjfields at uiuc.edu
Mon Aug 14 11:26:48 EDT 2006


Hilmar,

I originally mentioned using this branch as a possibility, which no  
one seemed to oppose at the time (I also mentioned using Bugzilla,  
dropping in patches or modules as code enhancements).  I felt it  
would be a way for everyone to try it out on different OS's (one of  
Sendu's main concerns) and still have the ability to update back to  
the main branch if it doesn't work.

I agree with your assessment of only using this for API changes,  
etc.  But it's my opinion Sendu is using it for the right purpose  
(he, and we, aren't sure whether these will work at all or to what  
degree they will work).  So far everything looks very promising,  
however we don't know how these changes will affect SearchIO and its  
API, esp. his and Aaron's proposed 'pull' parser using file pointers  
(which I quite like the idea of; something to think about for SeqIO  
if it pans out).  At least his code is available to be tested now!

If you think we should just have various experimental branches for  
each 'experiment' that could have API issues or not work, that would  
work for me as well.

I originally thought it would be a good idea to keep code in the  
experimental branch in line with bioperl-live because it would allow  
it's continued use as a testing ground, up to 1.6 and beyond if  
needed.  Otherwise I could see an experimental branch stagnating over  
time without bug fixes, code updates, etc.  Then half-baked  
'experiments' that didn't work would be tossed, ones that did would  
be merged back or committed to MAIN.  Of course, the long-term danger  
with a single experimental branch would be having too many  
experiments running on the same branch, which could make merging back  
a possible issue

Chris


On Aug 14, 2006, at 9:35 AM, Hilmar Lapp wrote:

> Don't complicate things more than they need to, please. The  
> experimental branch should be solely for things for which you  
> aren't sure whether they are going to work at all, and for API  
> changes for which the consequences across the board may be  
> significant and difficult to fully anticipate (e.g. the late- 
> breaking and still haunting SeqFeatureI changes before 1.5.0 should  
> have gone on an experimental branch first to see how they will  
> behave).
>
> It should never be a requirement to merge commits from the main  
> trunk to an experimental branch (or commit twice for developers).
>
> Quite frankly the changes Sendu described for me wouldn't have  
> warranted an experimental branch, they didn't sound like changing  
> API signature or API behavior.
>
> 	-hilmar
>
> On Aug 14, 2006, at 10:01 AM, Chris Fields wrote:
>
>> Sendu,
>>
>> Sounds good.  We need to make sure that commits to bioperl-live also
>> get committed to the experimental branch, correct?  Or at lease make
>> sure bioperl-live commits are merged into experimental (and not vice-
>> versa)?
>>
>> Chris
>>
>>
>> On Aug 14, 2006, at 6:43 AM, Sendu Bala wrote:
>>
>>> Chris Fields wrote:
>>>> Here's a couple of  suggestions to get around that if you want  
>>>> to get
>>>> the code out there for testing:
>>>>
>>>> Could this be CVS-tagged to an experimental bioperl branch instead?
>>>> It could be merged back to the main branch once everybody gets  
>>>> to try
>>>> it out, and you could commit changes to the branch (tests, scripts,
>>>> etc) along the way based on suggestions.  Think of this as a test-
>>>> drive for a new Bioperl release.
>>>
>>> I have created branch 'branch-experimental' and committed the
>>> changes there.
>>>
>>> Please test by checking out the experimental branch:
>>> cvs co -d experimental -r branch-experimental bioperl-live
>>>
>>> I'll probably end up writing a new pull/chunk parser for BLAST, but
>>> these changes will still speed up the other SearchIO modules. So  
>>> test
>>> the speed-up on different kinds of report as well.
>>>
>>>
>>> The experimental branch should be used for trying out major
>>> implementation changes that have the potential to break important  
>>> and
>>> substantial parts of bioperl. Everything else should continue to be
>>> committed to HEAD until the 1.6 branch emerges (sometime next year).
>>> _______________________________________________
>>> Bioperl-l mailing list
>>> Bioperl-l at lists.open-bio.org
>>> http://lists.open-bio.org/mailman/listinfo/bioperl-l
>>
>> Christopher Fields
>> Postdoctoral Researcher
>> Lab of Dr. Robert Switzer
>> Dept of Biochemistry
>> University of Illinois Urbana-Champaign
>>
>>
>>
>> _______________________________________________
>> Bioperl-l mailing list
>> Bioperl-l at lists.open-bio.org
>> http://lists.open-bio.org/mailman/listinfo/bioperl-l
>>
>
> -- 
> ===========================================================
> : Hilmar Lapp  -:-  Durham, NC  -:-  hlapp at gmx dot net :
> ===========================================================
>
>
>
>
>

Christopher Fields
Postdoctoral Researcher
Lab of Dr. Robert Switzer
Dept of Biochemistry
University of Illinois Urbana-Champaign





More information about the Bioperl-l mailing list