Bug#906222: gdal: undeterministically broken python packages

[Sebastiaan Couwenberg]

Control: tags -1 unreproducible moreinfo

On 8/15/18 5:38 PM, Sebastiaan Couwenberg wrote:
> On 08/15/2018 05:15 PM, Niko Tyni wrote:
>> I haven't located the cause yet; diffing build logs makes me wonder if
>> the 'generate' target of swig/python/GNUmakefile is getting erroneously
>> run in parallel with targets compiling the files that 'generate' modifies.
>> So maybe it's just a case of fragility on parallel builds?
> 
> This is most probable.
> 
> A parallel building issue was fixed in 2.3.1, see:
> 
>  https://github.com/osgeo/gdal/issues/707
> 
> There are likely to be more.
> 
> Since building gdal for more than one Python version takes a long time,
> I'm not in favour of disabling parallel builds.

So far I've not been able to reproduce this issue.

How massively parallel are your builds?

Mine are with "parallel=3", and the reproducible builds use 16.

Since my number of processes is less than the number of files to be
modified in the generate target, that may explain why those succeed.

We could remove the parallel option for the generate target only, that
may resolve this issue. That's mostly speculation, as I'm not intimately
familiar with the parallel make implementation.

Kind Regards,

Bas