Bug#687563: RFS: opengrm-ngram/1.0.3-1 [ITP] -- opengrm n-gram, library

[Giulio Paci <giuliopaci@gmail.com>]

Il 27/02/2013 13:27, Jakub Wilk ha scritto:
> (Sorry for the late reply. Somehow your mail fell through the cracks...)
> 
> * Giulio Paci <giuliopaci@gmail.com>, 2013-01-13, 13:56:
>>> Would it be possible to exclude binary files from the being analysed by licensecheck?
>> Done.
> 
> Now debian/source/include-binaries can be dropped.
> 
>>> Is there a reason you explicitly enable building static libraries?
>> Just to mimic openfst.
>> But I have to say that openfst upstream suggests static libraries for performance reasons, while opengrm-ngram upstream suggests no static libraries for compilation speed
>> reasons.
>> Do you prefer if I drop the static libraries?
> 
> If I were the maintainer, I wouldn't bother to enable them.
> But I'm okay with keeping them, too.

I prefer to keep them.

> Do you know how the files in src/testdata/* were generated? I wonder if we have the full source for it.

No I do not know. After manual inspection of the test scripts I can say that I do not know how to generate:
1) testdata/earnest.cat, testdata/earnest.det, testdata/earnest.fst, testdata/earnest.min (unused)
2) testdata/earnest.txt (used as input of ngramsymbols_test.sh and I guess the source of all the earnest.* files)
3) testdata/earnest.det.far, testdata/earnest.fst.far, testdata/earnest.min.far (used as input of ngramcount_test.sh)
4) testdata/init.randcorpus.0.mod, testdata/init.randcorpus.1.mod, testdata/init.randcorpus.2.mod, testdata/init.randcorpus.3.mod (used as input of ngramrand_test.sh)

All the other file have a generation command line example in test scripts:
testdata/earnest.far : ngramcount_test.sh
testdata/earnest.cnts : ngramcount_test.sh
testdata/earnest.syms : ngramsymbols_test.sh
testdata/earnest-count_prune.pru : ngramshrink_test.sh
testdata/earnest-relative_entropy.pru : ngramshrink_test.sh
testdata/earnest-seymore.pru : ngramshrink_test.sh
testdata/earnest.randgen.far : ngamrandgen_test.sh
testdata/earnest.arpa : ngramprint_test.sh
testdata/earnest.cnt.print : ngramprint_test.sh
testdata/earnest.perp : ngramperplexity_test.sh
testdata/earnest.mrg : ngrammerge_test.sh
testdata/earnest.mod : ngrammake_test.sh
testdata/earnest-absolute.mod : ngrammake_test.sh
testdata/earnest-katz.mod : ngrammake_test.sh
testdata/earnest-kneser_ney.mod : ngrammake_test.sh
testdata/earnest-witten_bell.mod : ngrammake_test.sh
testdata/earnest.info : ngraminfo_test.sh
testdata/earnest-det.cnts : ngramcount_test.sh
testdata/earnest-fst.cnts : ngramcount_test.sh
testdata/earnest-min.cnts : ngramcount_test.sh
testdata/earnest.randgen.apply.far : ngramapply_test.sh

I sent an email upstream to ask about the other files, but I guess we can suppose that we have all the reasonable sources for those files.

Bests,
	Giulio.