Re: Use gen-libm-test.py to generate ulps table for manual

[Carlos O'Donell <carlos at redhat dot com>]

On 08/13/2018 04:12 PM, Florian Weimer wrote:
> * Carlos O'Donell:
> 
>> On 08/13/2018 04:04 PM, Florian Weimer wrote:
>>> * Carlos O'Donell:
>>>
>>>> I'm impressed, you've done a really good job keeping this minimal, but what's
>>>> the problem with including python here?
>>>
>>> It would need a special bootstrap build, reducing external
>>> dependencies to the bare minimum.  A typical distribution build of
>>> Python pulls in X, Java and perhaps more.
>>  
>> You still haven't answered the "Why?"
>>
>> Why does your bootstrap need to be smaller?
>>
>> What is wrong with "typical?"
> 
> When we did this in Fedora, the bootstrap set was pretty much the
> whole distribution, so fully bootstrapped builds are extremely rare
> and generally happen only during architecture bringup, and often not
> even then.

It is a question of interest really. Fedora has the technical ability
to deliver a bootstrap that happens routinely, but I don't see the
motivation to do so. Only mass rebuilds happen routinely when Fedora
makes a release (don't have bootstrap issues).
 
> Andreas' setup is very good for detecting certain types of build
> issues, and I wouldn't want to lose that.

I assume you are arguing that a full bootstrap would take much longer
than the set of packages Andreas identifies. Please make such assertions
clear. I don't know if Andreas has automation to run the builds in
parallel across lots of machines and so it doesn't take that long to
add more packages.

The honest truth is that I'd like to see Python usable in the build
process for glibc, so I think we'd have to bite the bullet on a minimal
python at some point. $0.02.

-- 
Cheers,
Carlos.