This is the mail archive of the cygwin@cygwin.com mailing list for the Cygwin project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: [avail for test] libtool-devel-20030121-1


Ralf Habacker wrote:

1) x86 DLL (and EXE)
2) x86 archive static
3) x86 archive import
4) other

Do NOT try to combine 2) and 3) -- which is what all of your suggestions
in this email would do.

[1]
What about extending the 'file' tool, which supports already 1, (2 combined with
3) and 4 ?
You have written (and I have checked this too) that 'file's execution time is
independend from the object file size, so this would speedup this detecion.
Now that's an idea, but we'd need to get it into the "real" file distribution -- otherwise, cross-compile builds won't work.

I've taken a look into the file source and have seen the only thing I don't know
is how to define a rule for a string with a variable fileoffset, because
'_dll_iname' is, as far as i can see, one of major identifier of import
libraries.
Yep. Believe it or not, I *did* look at this idea early on -- and you've hit on the reason I didn't pursue it.

For an example how to add support with fixed file offset
/usr/lib/libintl.dll.a):

$ diff -ubB archive.orig archive
--- MagDir/archive.orig        2003-02-10 09:01:41.000000000 +0100
+++ MagDir/archive     2003-02-10 09:00:36.000000000 +0100
@@ -79,6 +79,8 @@
 >0     beshort         3               - shared library module
 >0     beshort         4               - debug break-pointed module
 >0     beshort         5               - absolute code program module
+>866   string          _dll_iname      import library
+
 0      string          \<ar>           System V Release 1 ar archive
 0      string          =<ar>           archive
 #
Sure -- but as you say, that only works for fixed offsets. Now, **most** import libs will need only the fixed offset, especially if you hunt for _dll_iname instead of " I " in $NM's output as I was doing. This is a very good idea.

But, you'd still need to check all "static libs" to see if they are really import libs after all. The good news it that we expect this to happen only rarely: when an import lib is a "hybrid" lib with static objs "in front" --> the modified 'file' test incorrectly identifies it as a static archive.

Question: for "normal" import libs (that is, excluding the hybrids like libcygwin.a), does your version work always? Or does _dll_iname 'float around' even within otherwise normal import libs?

Downside: we can't use this idea in libtool-1.5. We have to wait until after your change (after suitable debugging, of course) makes it into not just OUR file distribution, but the "real" file distribution.

Summary: sounds good. Please pursue, and once your modification has been accepted into 'file' and the upgrade has reasonable penetration on non-cygwin platforms, then we can *really* speed up win32_libid() -- maybe by libtool-1.5.1 or .2. But for now, I think we'll have to stick with the moderate speedup that my proposal allows.

--Chuck



--
Unsubscribe info: http://cygwin.com/ml/#unsubscribe-simple
Bug reporting: http://cygwin.com/bugs.html
Documentation: http://cygwin.com/docs.html
FAQ: http://cygwin.com/faq/


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]