Re: Bogus reports from coverage.postgresql.org

On 2019-06-01 19:10, Tom Lane wrote:
> Alvaro Herrera <alvherre(at)2ndquadrant(dot)com> writes:
>> On 2019-May-30, Alvaro Herrera wrote:
>>> On 2018-Mar-12, Peter Eisentraut wrote:
>>>> --enable-coverage disables ccache, so that's not going to be it.
>
>> I think I found the answer in this commit:
>> https://github.com/ccache/ccache/commit/02d3f078bd2495b8db2264ae0b2c692b4c5ba1bd
>> "Add support for coverage (compiling for gcov)"
>> committed for ccache 3.2.2.
>
>> So I think we could remove that stuff now ...
>
> I'd be a bit sad if we did, because where I usually do coverage checking
> is my RHEL6 box, which has
>
> $ ccache --version
> ccache version 3.1.6
>
> I'll probably bite the bullet and upgrade that box sometime in the
> next year or so, but it's not really a near-term project. Is there
> any big hurry for having ccache for coverage builds?

I happened to come across this again. I think we should make the change
Álvaro was proposing. The fix in ccache is 5 years old now.

For older ccache versions, you can always set CCACHE_DISABLE yourself.
Doing it the other way around is hard or impossible with the current setup.

--
Peter Eisentraut http://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services