What is needed for a new release:
- a working installer
- working plx conversion
Børre has called the company he had in mind, and discussed the installer issue with them.
We'll get an answer for whether they can fix it, and to what price, tomorrow.
- get a proper quote from the Tromsø company regarding a fixed installer/fixed
The PLX conversion now seems stable. The full speller behaves the same as the test speller in the tests we have.
Tomi did the regression test. The words contained in the test bench are accepted by both spellers.
I ran regression test with test speller. sme/polderland/regression.txt in general speller testing.
The PLX conversion test bench uses the following files:
Tomi is adding these tests to the proofingtools testing makefile.
Test results for running the words-*.txt files through the spellers:
- words-reject.txt + test speller: Rejects the same set as the full speller
- words-reject.txt + full speller: Rejects the same set as the test speller
- words-accept.txt + test speller: accpets the same set as the full speller, except "as" (due to a regex failure)
- words-accept.txt + full speller: accpets the same set as the test speller
Conclusion: the PLX conversion IS predictable and stable for the data we have.
- fix the PLX codes for the two reject strings presently accepted (Tomi)
- fix the as problem in the regular expression for test data (Tomi)
- work on test bench improvements (Tomi)
Meeting every afternoon, solving one issue at a time until we have a release candidate.