Closed Halbaroth closed 3 weeks ago
Looks good to me. What are the changes to
unzip.ae.zip
andunzip.smt2.zip
and why are they needed?
They did not contain the same test (up to translation). We only want to check that Alt-Ergo can unzip archives and recognize the right format based on the name of the archive. It would be annoying that one of them fails because the zipped test fails.
Looks good to me. What are the changes to
unzip.ae.zip
andunzip.smt2.zip
and why are they needed?They did not contain the same test (up to translation). We only want to check that Alt-Ergo can unzip archives and recognize the right format based on the name of the archive. It would be annoying that one of them fails because the zipped test fails.
Ah yes this is the issue that one is zipped with compression and not the other. We should keep one test for compressed zip and one test for uncompressed zip however. Can we instead rename them to be explicit (e.g. unzip_compressed.ae.zip
and unzip_uncompressed.smt2.zip
, or the opposite, I don't remember which is which)?
Now, there are both uncompressed:
file unzip.ae.zip
unzip.ae.zip: Zip archive data, at least v1.0 to extract, compression method=store
and
file unzip.smt2.zip
unzip.smt2.zip: Zip archive data, at least v1.0 to extract, compression method=store
I guess that zip
chooses automatically the compression method based on the size of the file. Compressing a tiny file will produce a table larger than the file itself, so zip
chooses the store
method.
If you want, I can force the method with the option -Z
, but I think it does not matter.
Please do keep a test with a compressed zip file. It is a use case we want to support so it should be in the test suite.
Done. The option -Z
cannot force zip
to compress file. So I added a padding comment in both tests to ensure that they will be compressed.
The module
My_Zip
contains only function which is called once inSolving_loop
. This commit cleans the code and move it inSolving_loop
directly. Technically,Solving_loop
is a part of the binary, but as everyone knows, most of the code of this module will be push into the library after we design a good API.This PR replaces #1248