#dev | Logs for 2023-06-27
« return
[16:57:14] <kolie> Proposal to make the makefile best effort and Docker the preferred build system going forward. Willing to hear from anyone with a commit into rehash because that's who this effects and deals with that decision. Reasoning on this is - the primary and secondary platforms rehash is used for are going this route and the makefile is becoming baggage to those efforts. It takes 15 minutes to build with the makefile - and working in docker you don't
[16:57:14] <kolie> get that 15 minute build everytime. Current work on a very small feature just lost a day to make the makefile place nice ( https://pasteboard.co - "works on my laptop where i did the work and not elsewhere" ) doing solved headbanging shit because of mixed environments.
[16:57:16] <systemd> ^ 03Pasteboard - Uploaded Image
[16:57:33] <kolie> mechanicjay,
[16:58:19] <kolie> If not - we likely need someone willing to maintain the makefile.
[16:58:35] * mechanicjay raises hand
[16:58:50] <mechanicjay> we shouldn't perscribe the way the software is run
[17:08:54] <kolie> I agree with that in principal. I'd like to know more about that as it applies here. For context a few things come to my mind 1) Who is running rehash and how are they running it 2) Does downgrading the makefile to "best-effort" impact that 3) Do we not already make many decisions on "how the software is run" by means of tech stack and versioning choices 4) Does supporting the makefile in prinicples - does that benefit outweigh the cost it
[17:08:54] <kolie> induces on future development. 5) If someone has an itch to scratch - my proposal would be that they take charge of the makefile if the primary use case for rehash makes it redundant and unfeasible for continued primary support
[17:40:36] <kolie> As it is - NCommander will be testing the makefile and if it works - it's getting commited along with the docker changes. If that works else where is the problem with makefiles and trying to make sure there is repeatability going forward.
[19:07:21] <mechanicjay> 1) We don't know, nor do we care
[19:07:21] <mechanicjay> 2) Unknowable
[19:07:21] <mechanicjay> 3) No -- we test that the software runs with certain versions of dependencies / prereq -- how and where those are supplied and satisfied shouldn't be our concern
[19:07:21] <mechanicjay> 4) I see a big downside to veryifying that the software will only run in docker -- it ties us unecessarily to a particular abstraction technology.
[19:07:21] <mechanicjay> 5) This is an issue with having a opensource project (rehash) have only one organization contributing.
[19:08:03] <mechanicjay> For myself, I'm going to continue hacking on rehash on hardware without containerization, so if nothing else, I'll be maintaining the Makefile for my own purposes.
[21:23:34] <kolie> re 5) if there was more active participants in the project then the makefile issue becomes moot. a code/steering committee would address Who uses it (#1) - how they use it (#2) and go on that. As there is no such structure in place for the code or its repo - it falls on I guess us here to try and work out what that is. Of course we should know and care who uses the repo and how! Knowing the use cases is very important in how you direct effort
[21:23:34] <kolie> in a project.
[21:24:36] <kolie> Upstream right now - I guess the PBC - will work in docker and update the makefile accordingly. We don't rely on the makefile in staging or production. As long as that's understood there wont be any suprises.
[21:25:59] <kolie> I wish rehash had a more active contributors list - it is what it is. SN may be the only installation out of dev workstations - that's entirely likely.
[21:58:58] <kolie> 840s into my makefile run. it's still done about ... 9 of the cpans.
[22:06:05] <kolie> 1300s in, down to Schedule::Cron in the cpan.
[22:19:35] <kolie> 28 minute flat build.
[23:12:13] * Bytram recalls that rehash was build on the source provided by /. and one of the principles that SoylentNews was build on that legacy and make it available to others just like we received. "Pay it forward", if you will.
[23:20:29] * Bytram notices there is something after "3)"
[23:23:21] <Bytram> I believe there had been other interrelations that were based on our code (ISTR Japan)