Re: [PATCH 2 of 2] cc/gcc: 4.4.5, add arm-softfloat-libgcc patch from buildroot

[Bryan Hundven <bryanhundven at gmail dot com>]

On Wed, Nov 10, 2010 at 1:52 PM, Arnaud Lacombe <lacombar@gmail.com> wrote:
> Hi,
>
> On Wed, Nov 10, 2010 at 4:47 PM, M P <buserror@gmail.com> wrote:
>> This patch makes ARM toolchains works again /for crosstool-ng/ -- if
>> it was adapted from a previous version that had the same behaviour and
>> fixed the same issue, what's the problem really ?
>>
>> SOME people are more concerned about getting stuff done than the way
>> it is done, I have absolutely no interest of getting a "proper fix"
>> from gcc guys that will appear with gcc 4.5.5 in 8 months time (or
>> whatever), I'm more concerned about getting gcc working on my platform
>> for my particular purpose, "today" if possible. Thanks to Bryan who
>> had been following the issue, I'm not staring at a non working
>> toolchain for the foreseeable future.
>>
>> It certainly makes sense to me, as my build now works...
>>
> Until you spend a month tracking down a compiler issue because someone
> pay more care about buildability than correctness, which compilers are
> all about.

I can definitely understand where you are coming from, I believe that
the problem is with your approach.

> If buildroot people had taken this to gcc guys, we would not even have
> this discussion.

So instead of trying to be helpful and having a constructive
conversation about this change, You decided to pull out the "insult
gun" and start firing.

Good Job.

> Â- Arnaud

To answer your question. It's a little bit more then 'patch -p1 < blah'.

I did take time and _look_ at the code that the patch would change.

With the limited resources I have received from Mychael (Mychel?), I
made an assumption that the code path affected would be for arm only.

I also used the power of the internet to look for anyone else having
similar issues, and if they were able to solve it, follow their lead.

I then made the suggestion to Mychael to try the patch (and that it
wasn't likely that it would work), and in the end it did work for him.
"Work" in this case, does mean build. It may not be in all
correctness, and so I followed up his response by running my own build
and running the test-suite on one of my arm boards. And it didn't
change my results at all (as I am building for xscale, and don't have
his issue).

I then sent in the patch. Hoping to get a conversation about it,
rather then getting insulted.

-Bryan

--
For unsubscribe information see http://sourceware.org/lists.html#faq