tech-toolchain archive

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index][Old Index]

Re: make: avoid breaking backwards compatability

>On Thu, Sep 04, 2014 at 04:32:31PM -0700, Simon J. Gerraty wrote:
>> target0 is the fiding some dir/foo.c via .PATH even when foo.c is
>> present in '.'
>This behavior would be a horrendous bug!  If the manual page should be
>trusted (which I've come to suspect lately...):

Sorry, but that is the way our make has worked for approximately ever.
This is what a number of people have been pointing out to you.

>And looking at your test case, I see that you probably meant to say:

I don't particularly care how you describe it, the test cases simply
show the historical behavior which is currently broken.

>Anyway, I will never re-implement this.  If you want this behavior, you
>either revert the changes or do it yourself.

Thanks for clarifying.

>> Stupid or not, in each case the real makefiles that rely on this
>> behavior have worked with every version of bmake posted for the last
>> 15
>> years.  They should contine to work.
>Are you really saying that we should allow ourselves to be held hostage
>by the past?  

To some extent yes.

>Because what you're saying means that no changes should
>ever be made.  

Rubbish.  We've added plenty of features/functionality to our make over
the years without breaking existing makefiles.
I don't see that we should depart from that practice.

Home | Main Index | Thread Index | Old Index