Give a pragmatic solution to suppressing noisy output

[steppi]

This PR replaces the broken means of suppressing noisy output from #150 with a pragmatic hack. (Note that although @Carreau submitted #150, I was the one who came up with this broken solution, and he was only implementing my idea).

The pragmatic hack I've made is to offer an option jupyterlite_silence in config.py. Setting this to True causes the stderr and stout from the call to jupyterlite build to be directed to dev/null. I've verified that this works for suppressing output, and that the notebooks still work. I have not made jupyterlite_silence=True the default because it's a drastic step which suppresses even warnings and errors.

I think it's reasonable to go with this hack, at least for the time being, until the necessary changes can be made in jupyterlite removing unguarded print statements and making doit for tasks configurable.

Many apologies for jumping the gun and making a broken release. I will be sure to learn my lesson from this. I will need help revoking the bad release from PyPI.

cc @Carreau, @martinRenou, @jtpio @matthewfeickert

[matthewfeickert]

Thanks very much for being so fast and responsive on this @steppi! Very much appreciated. 🙏

Many apologies for jumping the gun and making a broken release. I will be sure to learn my lesson from this.

I think we've all been there before, so no worries. :)

[Carreau]

I think we should check the recur code of subprocess run (CompletedProcess(args=..., returncode=1=...)) and fail if it is Non-zero (i'll check).

I also think that we might want to have stdout/err be PIPE and print them anyway if the return code is non-zero

[steppi]

Good idea.

[martinRenou]

Many apologies for jumping the gun and making a broken release

No worries! I stopped counting the number of broken releases I made 🙈

[steppi]

I've made it so stdout and stderr of jupyterlite build is still printed if the command fails following @Carreau's suggestion. I've also made jupyterlite_silence = True the default, following the suggestion of @rgommers here: scipy/scipy#20277 (comment).

[steppi]

Many apologies for jumping the gun and making a broken release

No worries! I stopped counting the number of broken releases I made 🙈

Thanks. That makes me feel better.

[steppi]

@Carreau, @martinRenou, @jtpio. Can we merge this and make a release? It would fix a broken build in one of SciPy’s CI jobs.

[Carreau]

Ok, let's merge this and try. It's ok to do broken releases, it happens.

[steppi]

Thanks @Carreau!

[jtpio]

Yes, thanks all for working on this!