@mattyvau Hello again.
Yes we have quite a substantial amount of technical debt and Python2 is one a major element of this. But the build system we have used for the entire projects history is I believe now no longer maintained and I think is python2 only so we have to find and implement an alternative. I’m hoping to get our Stable openSUSE version out first with near enough feature parity to our CentOS variant so we can at least get folks onto an upstream supported btrfs before we tackle these fairly massive system wide changes in the testing channel as that is what it’s for.
Any contributions you have on our options are welcome. I’m open to all suggestions but these will be massive changes across the entire project so stuff will likely be broken for a while. Another closely related issue is our Django version. We are hoping to jump to much later versions of both Python and Django to head off such things happening again in the future.
Do you run a source install by chance, and if so on which disto?
We have begun the move to Python3 but only in baby steps, ie importing the future library etc. But this is going to be a painful exercise that will likely have to be my main and possibly only project for quite a while. And Django updates are also frought with difficulty. But again I want to be able to offer folks an upstream supported btrfs prior to this with at least workable openSUSE builds.
We also now have issues with building in Tumbleweed so if you are building from source it would be a benefit to have some input on that front as I’m quite keen of offering both as we go forward. In that case I’m pretty sure it’s down to a Python update that breaks compatibility with one of our older libraries so I’m hoping a jump in that library will sort this. But again many of these dependencies, such as the nieve dependabot is always suggesting would simply break if we jump without first upgrading everything else, such as the Python2 to Python3 move and it’s consequent Django move. We certainly have our work cut out so do chip in if you have the time / expertise. Next most pressing issue it the Tumbleweed failure to build. This is important as it has been a really usefull indicator of what’s coming down the line as it were with Leap 15.x. We had for example a yum failure first, very recently, in Tumbleweed and that prompted us to move to dnf-yum. This same move was then shortly mirrored in Leap15.1. So I’m super pleased to have such early signals and the next one appears to be why we now fail to build in Tumbleweed.
Thanks for the heads up and yes we have seen a few messages of this type for quite a while but we have also had really old versions of btrfs for quite a while. So getting a working openSUSE build that can server our supporters/users while we grapple with the many ageing dependencies in the recently re-launched, openSUSE only, testing channel would be great. But we can’t do that until we have a more functional release in the testing channel so it can be promoted to the stable channel. Freeing up the testing channel for these bigger and more far reaching changes. I.e. updating everything and moving from a series of bolt on technologies to build in stuff, ie like Pythons async await and Some async stuff in Django 3 for example, but these newer ‘native’ technologies just didn’t exist at the time the project was started. And we haven’t had the resources to maintain dependency / core technology updates while doing our own development / bug fixes. So yes if you or others have the time to address these issues then super, I’m up for reviewing pull requests as they arrive, but we also have to get folks onto something usable in Stable for openSUSE before we load up the testing channel with all these changes.
Hope that helps, and thanks again for the heads up.