Re: fabric3

[Michel Albert]

On Tue, Jun 16, 2020 at 6:18 PM Mathias Ertl <mati@er.tl> wrote:
>
> Hi Jeff,
>
> On 6/15/20 9:40 PM, Jeff Forcier wrote:
> > [...] "were you using
> > fabric3 from PyPI? Fabric 1.15 is now effectively the same thing! come
> > on back!!"
> >

I've been following this thread a bit on the side and it is starting
to worry me a bit. I work at a fairly large company and about two
years ago we started migrating our CI and build chain away from Python
2, considering that it is EOL in 2020. We had already migrated most of
our code long before, but the build chain which relies heavily on
fabric was still in py2 (mainly because of fabric). We delayed it as
much as possible because fabric-2 never "felt right" for us because
the development seemed incredibly slow, and it was more cumbersome to
use than fabric-1, where the main issue was the inconsistent
"first-argument" being passed to tasks. Sometimes it was a connection,
sometimes it was a local context. This made it difficult to write
"shareable" functions easily. But that's another issue.

As 2020 approached we took the dive and migrated *everything* in the
build-chain to fabric-2 which was a huge effort. Especially
considering that every dev now needed to learn a new API, and rewrite
their fabfiles.

This "come on back" comment makes it sound like fabric-2 is being
abandoned and the development will now continue again on fabric-1. Is
that the case? So do we need to revert all our changes again we made
since 2018? What will happen with fabric-2?

I was honestly hoping that the EOL of Python2 would fuel further
development of fabric-2 as it currently seems quite dead (with the
last functional commit being from July 2019), and we might see more
activity on PRs.

This worries me as investing time again to downgrade would have a big
impact on our resources, and definitely a hard sell for the hierarchy,
as we already invested quite a lot of resources in the build chain.

Sorry for the rantish long text... but I'm panicking a bit now :|