Re: kofficetests validation

On Tuesday, November 09, 2010 16:30:30 pm Cyrille Berger Skott wrote:
> On Tuesday 09 November 2010, Jos van den Oever wrote:
> > To be clear, Chanis mail was not about the fact that koffice saves files
> > as invalid ODF in about 5% of the cases.
> 
> Ok.
> 
> > The question was, what to do about
> > the fact that many of the files in kofficetests seem to be invalid ODF.
> > Do we want to fix them? Chani is currently roundtripping files with
> > ODFKit from zip to flat back to zip and checking that the files do not
> > lose information or become invalid. Since many are invalid to begin with
> > the size of the useful pool of documents is smaller than it could be.
> 
> I am not sure if I follow you here... Why is loading an invalid document a
> problem ? (or testing loading such a document) Unless you are interested in
> a set of documents where you could check "load document_A; save
> document_B; test document_A == document_B" ?
> 
> But then we could just have two categories, valid document, invalid. I am
> just saying that because, my understanding is that kofficetests countains
> many "real world" document. If it is documents crafted to test a specific
> bug/feature, then making the document valid might be a good thing (unless
> you are interested in a bug that appears with invalid document), but
> transforming a "real world" document to be valid seems wrong to me.
Ok, I cannot say I disagree. Just that it would be nice to have many docs for 
the use-case you mention. Of course the case where we load invalid docs and 
save valid ones would be interesting, but that's reaching too far at the 
moment.

Cheers,
Jos

-- 
Jos van den Oever, software architect
+49 391 25 19 15 53
http://kogmbh.com/legal/