Stefano Zacchiroli

Just to recap for those who might not be aware of it, the expectation when using a CUDF solver is that upgrade scenarios will be preprocessed totally sorting all versions...

To be clear: preprocessing in the pipeline is not a goal per se (of course :)). But it's needed to address all the craziness in version numbering that exists out...

A couple more use cases from my personal experience (which I had to address with horrible sed/perl hacks…): 1) renaming accounts/restructuring the account hierarchy, without capture and without messing up...

About the CI failures, they seem all due to the fact "chardet" does not get installed in the CI environment. I'm not sure what's going on, but it seems unrelated...

> The CI uses the `requirements.txt` file in the repo to install dependencies. However, `requirements.txt` is incomplete as it at least misses the dependency on `chardet`. All this seems unrelated...

Both tar-ing and gzip-ing are not reproducible operations out of the box. (They of course *roundtrip*, but there is no guarantee that re-tar-ing or re-gzip-ing the same content will always...

Ah, just found our past discussion in #34 :-) That's great as ground work. The question on whether you'd be willing to include this into smart_importer (e.g., if it's contributed...

Avec l'augmentation exponentielle, le graphe actuel avec les cas totales devient de moins en moins lisible/utile. Ça serait effectivement super d'avoir le graphe avec les deltas journaliers. Merci d'avance!

OK for trips, that's my choice, and it's totally debatable. But any metadata that have a finite set of values is eligible for autocompletion. And you can't really say use...

On Tue, Jan 09, 2018 at 10:53:51AM +0000, Jakob Schnitzer wrote: > That seems to be to me the distinguishing feature between tags and > metadata: tags have a limited...