[RFC 00/17] Merge event loop implementations

[Pedro Alves]

On 2/14/20 2:21 AM, Tom Tromey wrote:
> Tom> However, my hope was that unifying on the gdb approach would be good
> Tom> enough.  And, while debugging on Windows this week, I happened (I forgot
> Tom> to look before starting, so it's kind of a goofy accident) to use this
> Tom> event unification branch.  I used gdb and also tried gdbserver, and as
> Tom> far as I can tell, everything worked fine.
> 
> I rebased this series recently.
> 
> I wasn't sure what the status really was.  I thought this at the time:
> 
> Tom> So, my view is that the current patch series is probably roughly ok.
> 
> ... meaning that gdb_fildes_t could be removed and replaced with plain
> "int", and nothing else would have to change.
> 

My latest thoughts were here:

  https://sourceware.org/ml/gdb-patches/2019-09/msg00555.html

"In conclusion, I suggest sweeping the issue under the rug for now
and cast SOCKET to int for this series, like GDB already does."

> I can re-test the series on Windows.  The updates due to rebasing were
> all mechanical, though, stuff like making sure the files were moved into
> the now-correct directories.
> 
> I thought maybe we could use gdb_select in gdbsupport, but now I see it
> reaches into the serial code, so maybe that isn't so easy after all.
> 
> Let me know what you think.

I can't find a response to my comments to patch 2 and 15, for example.
I'm not sure you saw them? 

Thanks,
Pedro Alves