[Unison-hackers] how far do we want to go in packaging?

Greg Troxel gdt at lexort.com
Sun Oct 25 09:52:17 EDT 2020


We had a question about how to install the CI-generated mac binaries,
and Roy suggested creating .app bundles.   Thoughts arising:

 - The main point of CI is CI.

 - We could go down the path of producing native packages for N
   packaging systems, where N tends to get very large.  In the limit
   this is crazy and impossible.

 - Many packaging systems tend to be behind, and the ocaml version
   problem means that one needs to use a portable packaging system with
   consitent builds across systems, but people dont'.

 - We do seem to have a lot of users that can't build from source or
   deal with this.

 - It seems someplace in the middle is warranted; I really don't want to
   go very far down this path.

 - That could be the missing instructions at
     https://github.com/bcpierce00/unison/wiki/CI-Binary-instructions

 - That could be making the CI binaries more package-y.

 - Part of the issue is the protocol break problems and fixing that is
   IMHO the most important thing to do.  (in progress, I know)


This is something I'm not personally interested in, as I use pkgsrc for
this (and maintain the unison entry in pkgsrc).  I'm happy to glance at
PRs, issue pleas for others to review, and hit merge if I think there's
consensus about the approach and that there's been adequate review.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 194 bytes
Desc: not available
URL: <http://LISTS.SEAS.UPENN.EDU/pipermail/unison-hackers/attachments/20201025/2729ff3c/attachment.asc>


More information about the Unison-hackers mailing list