gnu-arch-users
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Gnu-arch-users] Arch On Cygwin(Win32)


From: John Kinson
Subject: Re: [Gnu-arch-users] Arch On Cygwin(Win32)
Date: Tue, 18 Nov 2003 09:06:39 +0000
User-agent: Mozilla/5.0 (Windows; U; Windows NT 5.0; en-US; rv:1.5) Gecko/20031007

Robert Collins wrote:
What did you change them to? Did you find that methods on objects where
getting called with invalid 'this' pointers? And that the problem only
occured on stack based objects with a footprint > ~ 4K ?

If so, check the assembly, I strongly suspect it's a g++ issue, and have
a second hand report (good quality 2nd hand, but as there is no formal
test case..) that it is, and has been fixed in g++ 3.3.x. We should be
able to get a fix into cygwin gcc relatively easily, once we have a test
case.

Thanks for the hint, will look into it.

Actually, cygwin doesn't come close to the posix minumum. The MAX_PATH
define from the win32 headers applies to the paths used by cygwin. So
when you mount (for instance) c:\documents and settings\John as
/home/john/, you've lost ~20 characters silently. Add in server based
resources (mounted via \\ ... \ ....  and it gets worse.

Not much we can do about that then.

I just hacked successfully safe_mkdir to call CreateDirectoryW as a proof of concept, and this would seem *much* easier than the Cygwin patch... Any drawbacks to this approach?


It's ugly. It won't work. That should be enough.

With regard to the specific failure of creating long directories, actually it does work (very neatly). However...

Ugly I covered in my other email. As for working: you need to address
all the external tools:
~/.arch-params/hook
tar
gzip
bash

as a minimum.

Point taken.

Tell me that porting all of them to use win32 IO calls is easier than
fixing it in one place?

I can't answer that yet, but I'm willing to pursue it a little before ruling it out. I don't yet know whether *all* IO calls must be translated into W32 WCS native ones.

If, however, Cygwin is extended to support long paths, then are you in a position to roll the changes into an official Cygwin distribution? Would Cygwin be willing to accept such a change (with all the testing and support issues that come with it)?

If not, then arch will still face the significant obstacle of requiring a specially patched version of Cygwin in order to run.

JK
--
http://www.yellowradio.com/
What if the Hokey Cokey really _is_ what it's all about?





reply via email to

[Prev in Thread] Current Thread [Next in Thread]