Re: Potentially merging the transactional-memory branch into mainline.

[Diego Novillo <dnovillo at google dot com>]

On 11-11-01 11:23 , Jeff Law wrote:

> This stuff is fairly isolated in terms of what it touches and I'm sure
> if anything goes wrong, Aldy, Richard&  Torvald will be available to
> fix it.
>
> The request to merge came in before the end of stage1, I don't see a
> reason to delay things another 6-9 months.  This isn't like asking to
> pull in a whole new register allocator at the end of stage1 :-)
>
> Additionally, I believe we have a small window where we can position
> GCC to be the compiler of choice for those working with TM; waiting
> 6-9 months for GCC 4.8 will miss that window.

I agree as well.  I have not being following the TM work very closely, 
but I've been interested in TM for a Long Time.  Having it available in 
4.7 would be a huge benefit to GCC.

I don't think it really is a risk from a release standpoint.  TM is an 
optional component and should not affect standard codegen paths. 
Additionally, given that Aldy and Richard H. are involved in it, I'm 
sure they will be very quick in addressing any problems that crop up.

Aldy, Richard, is there a patchset or master patch I could read?


Diego.