[haiku-development] Re: AHCI regressions
- From: Axel Dörfler <axeld@xxxxxxxxxxxxxxxx>
- To: haiku-development@xxxxxxxxxxxxx
- Date: Wed, 9 Sep 2015 13:21:09 +0200
Hi Stephan,
Am 09.09.2015 um 13:09 schrieb Stephan Aßmus:
To be honest, I don't think it is good style to commit stuff which
causes boot failures.
That's a bit of an over-simplification: this is a hardware related
change; what works for others might not work for you, and vice versa.
Anarchos tested the patch on his hardware, I tested on Qemu. Since both
turned out to be successful, I committed his patch. It definitely didn't
look problematic to me which was obviously too optimistic.
This should be developed in a branch until it is
ready (or expected to be good enough), that is what they are for.
A branch usually does not help with testing on different hardware.
It's basically a choice of the development model: should trunk always be
stable or not?
Even after the fallout, I'd probably still do the same mistake again in
the future. What we miss is a relatively stable branch -- the nightly
builds aren't really helpful if you want is a *working* on-the-edge system.
Has
anybody already narrowed down the commit that is responsible? Even if
there is work going on to fix the fix, can this please be reverted for
the time being?
Since there have been some cleanup commits in the mean time, I'd rather
just disable the interrupt check again for the time being.
Bye,
Axel.
Other related posts: