Question about replacing a stub...
halkun2002 at yahoo.com
Mon Dec 29 00:47:52 CST 2003
Thanks for the reply...
--- "Gregory M. Turner" <gmturner007 at ameritech.net>
[A Bunch of really insightful stuff]
Implementing should best be left to the implemnters.
I'll leave the stub in the .spec file, leaving the
unimplemnted functions undocumented. This has some
advantages. First c2man,pl was thinking that the
function was actually implemnted and incorrectly
updating the stats. Second, when you have an HTML page
full of links to various API definitions, the stubs
start to really jump out at you.
But some issues do arise. One of them is a rather
intresting thing I found out about advapi32.dll. There
are 33 undocumented system fuctions,
SystemFunction001 through SystemFunction033. It has
come to my attention via an NT Crypto god that these
are actually encrypters/decrypters for various
ciphers. RC4, MD4, DES, ECB mode DES, and the like. It
would be kinda keen to put a "stcky note" somewhere as
to not lose gems of information like that. You can put
in roumors and other tales of undocumented intreague.
Then gain, I guess I could always push comments like
that out to the DLL document itself, or just snuggle
them in the code where they won't get parsed by c2man.
Doing the documentaion isn't that bad. A lot of the
handles, pointers, discriptors, and wotnot are used
over and over and over again. The hard bit is not
using MSDN, but that's tricky as all the other API
websites I've found blatently rip off MSDN anyway.
This coupled with the facts that the auto-genertaed
API docs are fightenly close to MSDN's API format, and
that you can only really document and API so many
ways, You find yourself in a box a little smaller than
confortable. I refer to MSDN as a refrence, but only
to make sure that I didn't make a copy of what they
said with my own words on accident.
Anyway I'm up waaaay too late now. I'm going to submit
patches on a dll-by-dll basis.
You have not heard the last of me........
More information about the wine-devel