Re: Recent checkins break 'make' on staging

[David Kastrup]

James Lowe <address@hidden> writes:

> Hello,
>
> On 03/02/16 19:26, David Kastrup wrote:
>> Paul Morris <address@hidden> writes:
>>
>>>> On Feb 3, 2016, at 1:38 PM, David Kastrup <address@hidden> wrote:
>>>>
>>>>> Can someone help? Or perhaps remove these entries from staging for the
>>>>> time being?
>>>> I'll back out issue 4752 and let you cater for further cleanup.  Holler
>>>> if further help is needed.
>>> Thanks David and James.  Sorry these patches are not working for some
>>> reason.  I’ll try to take a closer look soon.
>>>
>>> Any ideas on why make succeeded both for me and for James during
>>> testing / code review, but not now on staging?  (Also, the LilyPond
>>> that I built using make worked fine for me when I was testing out the
>>> new functionality.)  Seems odd...
>> Different GCC versions are a typical cause.  Not using
>> --disable-optimising (alternatively --enable-checking) when testing is
>> another.  Having forgotten to check in some new file also happens
>> regularly.
>>
>
> Thanks David, I've always use --disable-optimising for 'configure'
> when testing, do you know if I should be running --enable-checking 'as
> well' or 'instead' or both together or using the --enable-checking
> switch explicitly when running one of the other commands (i.e. make
> instead of configure)?

We should likely tell Patchy to use --enable-checking instead of
--disable-optimising.  At the current point of time, the latter implies
the former.  But since the final builds are made while optimizing, it
would make more sense to do the tests in that manner as well.  In my
experience, it's sort of a tossup with regard to build times: the build
itself takes longer, doing the testing is faster then and so is building
the docs.

-- 
David Kastrup