Author serhiy.storchaka
Recipients larry, nadeem.vawda, serhiy.storchaka
Date 2014-01-08.21:53:07
SpamBayes Score -1.0
Marked as misclassified Yes
Message-id <23548625.lTrPWFr8ZA@raxxla>
In-reply-to <1389213600.29.0.167206660992.issue20193@psf.upfronthosting.co.za>
Content
Here are completed patches for the _bz2 and _lzma modules (only constructors 
are not converted) and preliminary patch for the zlib module.

There are issues in the zlib module:

* Argument Clinic generates invalid code for zlib_Decompress_flush 
(issue20196).

* Argument Clinic can simulate optional positional parameters without default 
Python values only using groups, and resulting code differs from original and 
is too cumbersome (see code for zlib_compress, zlib_decompress, 
zlib_Decompress_flush).

* I have experimented in zlib_decompressobj, and I got good compact code, but 
wrong docstring ("decompressobj(wbits=None, zdict=None)"). Needed a way to 
specify optional parameters without default Python values. Correct signature 
should be something like "decompressobj([wbits], [zdict])".

* Previous approach is not applicable to compressobj because Py_buffer can't 
have default value. Seems as compressobj is not compatible with Argument 
Clinic.

* This is not specific to this patch, but pydoc produces wrong outputs for 
compressobj.

Perhaps there are other issues, this is only preliminary patch.

Larry, I don't want to scatter patches for these related modules and merge 
them with patches for totally unrelated modules. And I think it would be 
cumbersome to Nadeem Vawda to make reviews in such circumstances.
History
Date User Action Args
2014-01-08 21:53:07serhiy.storchakasetrecipients: + serhiy.storchaka, larry, nadeem.vawda
2014-01-08 21:53:07serhiy.storchakalinkissue20193 messages
2014-01-08 21:53:07serhiy.storchakacreate