Recent changes to pthread pre/postcreate and init_main_thread

Christopher Faylor
Thu Dec 11 16:47:00 GMT 2003

On Thu, Dec 11, 2003 at 12:36:04PM +0100, Thomas Pfaff wrote:
>while i can understand the reasons for your changes to the code i am not 
>happy whith these regarding the cancel event creation:
>Doing it in postcreate after a thread is already created is nonsense 

Hopefully, this is just a language difficulty here, otherwise I'd certainly
have a problem with someone calling a change I made "nonsense".

But in the spirit of this observation, let me point out that there were
multiple places in the thread code where the same thing was being done
in multiple places.  The debugging output in situations like this was
subtly different.  The error handling was often different.  To me, that
is an indication of a need for a function.

Then, on the other hand, we have functions which are two lines long,
called once.  To me, these call for consolidation.

I know that the code has had a checkered past and that it is probably
suffering a little bit from that so I haven't been unduly concerned.

As you've probably inferred, my goal in creating the cancel event in
postcreate was to use the same code for both the main thread and for a
created thread.  You're right that failure to create an event would
mean that a thread would be created without its cancel event.  I'd
ask you for suggestions on how to handle this but...

>Apart from this i will not longer maintain the pthreads stuff.

Ok.  Thanks very much for your contributions.

I assume that you will be unsubscribing from cygwin-developers?


More information about the Cygwin-developers mailing list