Em dom., 22 de set. de 2019 às 04:10, IOhannes m zmölnig <zmoelnig@iem.at> escreveu:
Am 21. September 2019 08:45:38 MESZ schrieb Alexandre Torres Porres <porres@gmail.com>:
I see that puredata-core has cyclone, but an old version, how can  include the latest cyclone?

Debian (and derivatives) ship "cyclone" for more than a decade. Because your fork of cyclone
- breaks functionality (starting with missing "cyclist")
- and the takeover was not very well coordinated
... i have not switched (yet) the Debian packages to your upstream.

I think the biggest issue for me is (still) the uncoordinated takeover.
if that is resolved, the Debian packages might (and i don't see why not) switch

Howdy. Not sure I get the takeover issue and how to solve it or not. I think breakage is more serious but we hope to solve both issues so debian packages can switch.

Regarding breakage, cyclone 0.3's goal wasn't to break stuff. If that happened, it should be reported as a regression bug. As for cyclist, we brought it up in https://lists.puredata.info/pipermail/pd-list/2016-10/116633.html - we questioned if anyone ever used and how it wasn't there for the binaries in alpha-57 and 0.2 beta1 (when we forked). It had been removed with [maxmode] and it wasn't our intention to take it out. We asked if and how we should bring it back as we didn't really know how to do it or how it would work, as it was part of [maxmode]. Anyway, the discussion didn't get anywhere and cyclist was kept out as we just didn't know what to do. But we can put it back, no problem. Help is needed though.

As for other known breakages, cyclone 0.4 announced how Capital Letter aliases were removed (for objecs like [MouseState]). The truth is that aliases weren't working for linux in cyclone 0.3 as a bug report informed us. I actually reckon it was a lazy attitude to remove support instead of trying to fix this (we found a rather complicated solution we didn't go for). And since there was a talk in the list how this could be phased out (katja mentioned that and so did cyclone 0.3), I thought we could go for it. Nonetheless, nowadays I also feel like we could take our time and try and fix this in an easy way to avoid breakages, so cyclone 0.5 could bring this back and just leave it there. Other breakage issues like the cyclist one can be solved.

So, bottom line is, we hope to deal with any uncoordinated issue or possible breakages. As for the coordination issue, I'd like to understand what it involves. As for the other issue, please report of other breakage issues if they are and you know of.

For the record, changes we made are in a changelog currently available with cyclone, find it here: https://github.com/porres/pd-cyclone/blob/master/documentation/extra_files/CHANGELOG.txt - all we did was fix bugs and update objects (plus add new objects and rewrite the documentation). Again, the goal was not to break stuff. Actually, we restored a change that had been made, which I consider a breakage: the non alphanumeric objects were restored into a 'cyclone' library binary as originally.

Cheers