Outstanding issues with current DLL?

Christopher Faylor cgf@redhat.com
Sun Mar 11 12:43:00 GMT 2001


On Sun, Mar 11, 2001 at 11:11:26PM +0300, Egor Duda wrote:
>Hi!
>
>Sunday, 11 March, 2001 Christopher Faylor cgf@redhat.com wrote:
>
>CF> Btw, the reason for the --enable-debugging switch is to build a cygwin1.dll
>CF> which timestamps all of its shared memory stuff.  This allows two versions
>CF> of cygwin to be running on the system at the same time.
>
>do   we   really   need   this?  suppose a some problem  reports about
>current  snapshots  we've  seen  here this week can be due  to several
>cygwin1.dll's.  at least freezes can be easily explained if we suppose
>such  possibility. when process 'exec's some cygwin program which sees
>other  cygwin1.dll,  it surely will break in an unpredictable way when
>doing fdtab.fixup_after_exec in dll_crt0_1()

Hopefully the version numbering in the exec header should catch this.
I'll bump it up, just to make sure.

>using  CYGWIN_TESTING  environment  variable was enough for me, and at
>least i always knew what i doing.

CYGWIN_TESTING is a relatively new addition by DJ which just adapted
code that had been in Cygwin for two or three years.

So, yes, we need this.  I've been using it for a long time and I
rely on it.

I didn't say that this was foolproof.  It is far from it.  If it was
foolproof it would be the default.  However, I find it invaluable for
testing.

If this is really the cause of people's problem reports, I'll be surprised.
I just fixed one of these and it certainly wasn't due to disparate cygwin
DLL versions.

cgf



More information about the Cygwin-developers mailing list