This is the mail archive of the cygwin-developers 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: Deprecating ntea


On Tue, Feb 27, 2007 at 05:35:26AM -0800, Brian Dessent wrote:
>Christopher Faylor wrote:
>> *cough* cygwin v2 *cough*
>
>So, when the major version number increases, we'd now have cygwin2.dll,
>which would have a different shared memory region ID, mount table
>registry key, etc... and so we'd have both cygwin1.dll and cygwin2.dll
>present in /usr/bin, until such time as all apps linked against
>cygwin1.dll got rebuilt from source (i.e. forever, or at least a
>reaaaaly long time)?  Would cygwin v1.x apps recognise v2.x apps and
>play nicely and vice versa?  And this would be a chance to drop a lot of
>cruft that is currently there just for backwards compatibility?
>
>I'm just trying to get an idea of what the logistics of such a bump
>would involve.

I raise the issue of cygwin2.dll with Corinna on a regular basis and
we eventually convince each other that it would be too much work.

What I envision is a backwards-incompatible DLL which would require
recompilation of applications.  It might be possible to write a
cygwin1.dll wrapper which called cygwin2.dll to avoid the need to
recompile absolutely everything - at least initially.

Off the top of my head things I'd like to fix would be:

signal mask
termios structure
environ passing
user_data passing from program to cygwin

I'm sure there are more...

cgf


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