relative IRI resolution in w3c tests about rdf4h

[robstewart57]

Relates to #17 .

[cordawyn]

I implemented W3C/*Test.hs such that they use a local dir (see suiteFilesDir) for base URI in order to resolve paths to the test files (mf:action and mf:result use relative paths). If the expected test results have that web path hardcoded, we can process those two (and similar) entries differently -- massage the resulting absolute URI in mfEntryToTest, where the file names are fed to parseFile, such that they represent local (relative) file names.

Ideally, we should take those absolute URIs as parser returns them and download the corresponding files from the web, but as usual, we have to make use of some dirty hacks to avoid impracticality ;-) There's also an option to implement some transparent caching of downloaded files, such that the files are downloaded only once and their local cached copies are reused as "responses" for consequent HTTP queries. But that sounds a bit too strained, doesn't it?

[robstewart57]

All W3C tests in rdf-tests now pass, so I assume this issue is now fixed.