[visit-developers] help with interface to test results

Cyrus D. Harrison cyrush at llnl.gov
Thu May 7 18:16:37 EDT 2009

Sure - just note the query interface won't likely be accessible from 
html without a webserver.
This isn't a problem, I still think the database will be very useful. 
Its' fine to not include the actual data in the database, but It may  
make sense to encode relative paths to the files as fields.


Mark Miller wrote:
> Hi Cyrus,
> So, I think what I am going to do is use sqlite to store test results in
> a database that will grow and grow over time. But, I think this database
> will contain 'rows' of information about each test and NOT actual test
> results (.png or .txt) files. So, this sqlite thingy can house results
> from many platforms each night and for many, many years of running
> without getting 'too big'.
> I think as Tom suggests, I'll have some other python processes that
> produce HTML pages (for most recently generated test results say going
> back 3-6 months) including all .png and .txt results.
> Later, we can add 'forms' to query info from the sqlite db and return
> results as Brad and others have suggested (e.g. show me all cases where
> NASTRAN failed on a Linux architecture in the last 2 months). But' we'll
> be hitting the limit of my web programming understanding then ;)
> Does this sound reasonable?
> Mark
> On Thu, 2009-05-07 at 12:27, Cyrus D. Harrison wrote:
>> Hi Mark,
>> Ideally - we would have access to a webserver that provides a language 
>> (PHP for example) for generating the html on the fly.
>> With this type of setup we could use html forms to connect to a (sql) 
>> database and generate dynamic content easily.
>> While none of this is hard, it certainly can be time consuming & we 
>> currently do not have access a suitable server (unless perhaps nersc 
>> provides php/database support?)
>> Another possible option is using xml & xslt - I use this (only for 
>> formating) for the build_visit testing, this is actually a client side 
>> deal.
>> I think freeing this info from just statically generated html pages 
>> would be very helpful, and it is something we should chat about more,
>> -Cyrus
>> PS: (visittesting.org + visittracker.org)
>> Mark Miller wrote:
>>> Hi All,
>>> Ok, I am really lacking in my knowledge of how web apps work. I
>>> understand browsing something like
>>> "file://   /alderaan.llnl.gov/visit/index.html" where a bunch of statically
>>> created and linked HTML files are visible. Thats basically what we
>>> currently do.
>>> I'd like to have a place anyone can go to look at test suite results
>>> from previous night, week, month, etc. I like people's suggestion of
>>> making a 'database' of test results and I think I can figure out how to
>>> do that. But, the part I don't understand is how I make a web interface
>>> to this 'stuff'. I assume I'll be spending the bulk of coding on a
>>> 'server' that will stand between developers and the test results
>>> database serving up stuff to you as you 'browse' it. I assume I cannot
>>> do this reasonably without coding a 'server'. Is that correct? I mean I
>>> can't or shouldn't do this as an 'applet' (I don't really know that that
>>> is) that runs in the client browser. I am thinking of using web.py to
>>> build the server.
>>> Am I at least on the right track here?
>>> Mark

More information about the visit-developers mailing list