Re: work in arm64 branch or on main branch?
Tim McNerney <[email protected]> Tue, 5 Mar 2024 12:41:57 -0500
| Newsgroups | gmane.lisp.openmcl.devel |
|---|---|
| Message-ID | <[email protected]> |
Linux distribution no, but I do have a PPC iMac with CCL installed that I am= willing to share as a server. How out of-date is the PPC CCL compiler? Can that be concisely described? --Tim > On Mar 4, 2024, at 18:55, Robert Munyer <[email protected]> wrote: >=20 > =EF=BB=BFOn 6 January 2024, R. Matthew Emerson wrote: >=20 >> As I do little bits of work on an ARM64 port, is there any killer >> benefit to keeping that work in a separate arm64 git branch? I was >> thinking that committing to the main branch would make visibility >> better, and as long as the main branch always builds, the overhead >> of a long-lived branch seems not worth it. >=20 > When I thought about this question last summer, I concluded that > the answer depends on the vintage of the code that's being modified. >=20 > If I were modifying code that's current as of CCL 1.12.2 > (e.g. the x86 compiler), I would put the changesets on trunk. >=20 > If I am modifying code that was last supported in CCL 1.10 > (e.g. the PPC compiler), I should make a new branch that starts > at the last 1.10 changeset, to avoid having 1.10 code and 1.12.2 > code being developed in the same branch at the same time. >=20 > (Another option would be to start by upgrading the PPC code to > 1.12.2, but I've had difficulty finding a Linux distribution that > can run ppccl64; PPC64 Linux distros have gone little-endian.) >=20 > P.S. I don't claim any copyright on my arm64 changesets, so the > files in my repo are IP of whoever owns the files in the main repo. >=20 > -- Robert Munyer > https://ccl-arm64-2023-07.srht.site