| 18 Dec 2025 |
 emily | (I also don't think one GCC build per target is that bad though) | 15:46:18 |
| emily | (if you don't want to rebuild a compiler per target my suggestion is to use LLVM :) ) | 15:46:26 |
 @rosssmyth:matrix.org | I would like to use llvm at work but that has its own challenges that I have not been able to solve yet. Mainly around libgcc and linker scripts. | 15:47:12 |
| @rosssmyth:matrix.org | and asm | 15:48:07 |
| @rosssmyth:matrix.org | Because Clang does not support the same asm as gcc, it supports mainly a subset of what gcc does | 15:48:37 |
| emily | do you target so many ARM versions (with so little builder resources) that building some GCCs is untenable? | 15:49:19 |
| emily | that said, multilib support wouldn't be a terrible thing | 15:49:27 |
| emily | it would certainly make K900 very happy, for i686 reasons | 15:49:32 |
| @rosssmyth:matrix.org | Mainly just two. | 15:50:13 |
| @rosssmyth:matrix.org | Once in a while a third one | 15:50:38 |
| @rosssmyth:matrix.org | And the builder resources & cache is just my laptop. Such is life as a software engineer at a hardware company. | 15:51:58 |
| emily | for prod builds too? 🫣 | 15:52:49 |
| @rosssmyth:matrix.org | I don't really mind building GCC on my laptop. The real issue is GCC on GHA. | 15:52:56 |
| @rosssmyth:matrix.org | Of course. | 15:53:03 |
| @rosssmyth:matrix.org | Working at a company as the newest hire in like 20 years has some funny things about using contemporary software engineering practices | 15:53:53 |
| @rosssmyth:matrix.org | I am one of two software engineers, and I am the only one who uses version control, the only one who cares even a little bit about being able to build our firmware outside of IDE tooling, ci, and other fun stuff | 15:54:59 |
| emily | can you give a rough estimate of how many countries' national security depend on the integrity of your laptop? or maybe planes' ability to remain aloft or medical devices' proper functioning? :D | 15:55:06 |
| emily | but yeah, big builds on GHA is a pain | 15:55:14 |
| emily | and I assume your company isn't up for paying for the beefier or third-party runners | 15:55:32 |
| @rosssmyth:matrix.org | I know we interact with the US military but I'm unsure what we do with them.
We have major contracts with the single company that does almost all testing of silicon wafers for all foundries
We have many contracts with various medical device manufactuers
No aerospace contracts (yet!)
And yeah, we just have the cheapest github plan at like $50/year | 15:56:53 |
| @rosssmyth:matrix.org | Also if you drive a Tesla Model X, we have hardware in the rear power seats | 15:57:09 |
| @rosssmyth:matrix.org | Though starting next year we will not as they are moving that stuff in-house iirc | 15:57:34 |
| @rosssmyth:matrix.org | We also supply motors for the McDonald's Ice Cream machine. | 15:57:51 |
| @rosssmyth:matrix.org | * I know we interact with the US military but I'm unsure what we do with them.
We have major contracts with the single company that does almost all testing of silicon wafers for all foundries (KLA)
We have many contracts with various medical device manufactuers
No aerospace contracts (yet!)
And yeah, we just have the cheapest github plan at like $50/year | 15:58:12 |
| emily | thankfully there are other components of the Tesla supply chain I am more concerned about, such as Tesla | 15:58:29 |
| @rosssmyth:matrix.org | true... | 15:58:36 |
| @rosssmyth:matrix.org | They suck to work with! | 15:58:44 |
| emily | haha, don't get fired for complaining about your clients on NixOS Matrix :) | 15:59:12 |
| emily | (aren't those the ones famous for always being broken, though?) | 15:59:28 |
| @rosssmyth:matrix.org | they are not our client anymore, the contract expired anyways | 15:59:35 |
