[2026-03-11T06:03:46Z] oh no, but recursive make is so good :( [2026-03-11T08:49:31Z] midfavila: that article is good to know pitfalls people did writing recursive makefiles [2026-03-11T08:50:00Z] the conclusion of the paper is wrong. Recursive makefiles are nice, but you have to deal with the problems that the paper raised [2026-03-11T08:50:16Z] scc uses recursive makefiles and it does not have any of these problems [2026-03-11T08:51:07Z] sad_plan: ok, I'll try kiss and see how far I go [2026-03-11T09:03:22Z] k0ga: scc is probably better written than lot of these project the article is refering to :p [2026-03-11T09:04:41Z] k0ga: yeah, you do that :D [2026-03-11T09:16:30Z] sad_plan: there was a lot of tradition about how to write recursive Makefiles [2026-03-11T09:16:43Z] and this article raised the problems [2026-03-11T09:16:59Z] for example, for loops to handle subdirectories [2026-03-11T09:17:40Z] in scc I just create rules for directories to enter in them, so they can be proccessed in parallel [2026-03-11T09:18:05Z] scc needs more testing in actual projects [2026-03-11T09:18:34Z] it has many small tests, but I am pretty sure it is full of bugs that I will only see trying more complex programs [2026-03-11T09:18:43Z] dwm is complex enough to raise problems [2026-03-11T09:19:20Z] sbase is also a good target, and we added a set of tests recently [2026-03-11T10:10:26Z] hm. how does scc fare in comparisson to cproc? ive used cproc for several stuff, and i know lots of stuff builds just fine with cproc. but it lacks a preprocessor, which isnt ideal... [2026-03-11T10:10:40Z] but I agree, the smaller compilers needs more testing so that they can improve [2026-03-11T10:12:03Z] k0ga: i new sbase added lots of tests recently, but didnt know they were scc based. appart from the make from scc which was added, due to some issues on bsd's make iirc [2026-03-11T10:12:56Z] nooooo, they were not scc based [2026-03-11T10:13:08Z] I meant that I can use sbase tests to test scc [2026-03-11T10:13:18Z] at this moment I know that scc can compile sbase [2026-03-11T10:13:28Z] but I don't know if the code generated is correct [2026-03-11T10:13:56Z] there is where sbase tests can help me, because if any test fails with scc but it works with gcc/clang then there is a bug in scc [2026-03-11T10:21:39Z] ah, right [2026-03-11T21:17:57Z] hi [2026-03-11T21:18:14Z] midfavila: do you own a reform, right ? [2026-03-11T21:19:24Z] yes [2026-03-11T21:19:33Z] why do you ask? [2026-03-11T21:20:05Z] I've got curious about the mnt reform next [2026-03-11T21:20:12Z] aah [2026-03-11T21:20:17Z] it looks interesting [2026-03-11T21:20:29Z] i like the way they split the mainboard up [2026-03-11T21:20:55Z] how difficult would be to port kiss to it ? Do you have an idea ? I don't know if you already tried something [2026-03-11T21:21:10Z] oh, i already crosscompiled the rootfs [2026-03-11T21:21:43Z] all you would have to do for a quick and dirty port is take that rootfs, rip out the debian rootfs from the disk image, and replace it with that one [2026-03-11T21:21:54Z] add in some of the device specific software if necessary like the kernel and there you go [2026-03-11T21:22:00Z] same thing i did for my uconsole [2026-03-11T21:22:15Z] i haven't actually sat down and ported kiss to the reform yet, but it's on the list [2026-03-11T21:23:31Z] hhhmmm.... of course with kiss I have compiled my kernel. [2026-03-11T21:23:53Z] but I have no idea what you mean with "crosscompiled the rootfs" [2026-03-11T21:24:09Z] like the kiss filesystem's default packages [2026-03-11T21:24:25Z] i built them by hand using an x86 system with a cross-compiler [2026-03-11T21:25:07Z] what is a cross-compiler ? [2026-03-11T21:25:58Z] it's a compiler that runs on one architecture, like x86, but produces binaries for another architecture, like arm [2026-03-11T21:28:58Z] ah ok [2026-03-11T21:30:15Z] maybe stupid question, but isn't it possible to boot the reform with an archlinux installer for example and then to compile the kernel on the reform ? [2026-03-11T21:32:00Z] yes, but my reform has an exceptionlly weak processor compared to even a 20yo desktop [2026-03-11T21:32:06Z] it uses an imx8mq [2026-03-11T21:32:27Z] the intial launch module with four low-power cores and 4gb of memory [2026-03-11T21:38:46Z] do I understand correctly ? You have a first model with an imx8mq [2026-03-11T21:39:16Z] and now the reform has an rk3588 [2026-03-11T21:40:00Z] yes [2026-03-11T21:40:03Z] that's correct [2026-03-11T21:40:39Z] i don't intend to upgrade my laptop's processor until there's a more open replacement [2026-03-11T21:40:54Z] the imx8mq remains, to my knowledge, the module that requires the fewest blobs for laptops [2026-03-11T21:41:05Z] ok [2026-03-11T21:41:06Z] (just to initialise memory iirc, after that it's all free software) [2026-03-11T21:41:40Z] i'm hoping to acquire an ls1028a once the desktop variant launches and then i'll be putting this mainboard in that housing [2026-03-11T21:41:55Z] with an old nouveau supported gpu [2026-03-11T21:43:35Z] I've just discovered they plan to launch a desktop variant. Do you know where I can find some information ? Probably somewhere in the news on their website [2026-03-11T21:45:44Z] it's just a housing [2026-03-11T21:45:53Z] but yeah they have a blog post [2026-03-11T21:46:22Z] https://mnt.re/media/reform_md/2026-02-28-feb-update.html [2026-03-11T21:46:24Z] "mnt station" [2026-03-11T21:48:06Z] thank you. [2026-03-11T21:49:03Z] let's assume kiss can be installed on an mnt reform. As I have seen, there is no fan to cool the cpu. [2026-03-11T21:49:53Z] none is necessary [2026-03-11T21:50:06Z] at the moment I have an intel core ultra 5 125h [2026-03-11T21:50:07Z] they're extremely low power and don't get hot [2026-03-11T21:50:11Z] a heatsink is sufficient [2026-03-11T21:50:19Z] hhhhmmm [2026-03-11T21:50:49Z] my uconsole doesn't have a fan either [2026-03-11T21:51:04Z] its rinky-dink little 1ghz riscv core doesn't get hot enough to need one [2026-03-11T21:51:09Z] it just dumps waste heat into the chassis [2026-03-11T21:51:29Z] I need about one hour to compile firefox for example. No risk of overheating, if I want to compile it ? [2026-03-11T21:51:46Z] and maybe damage something [2026-03-11T21:52:36Z] you're not building firefox on this [2026-03-11T21:52:37Z] lol [2026-03-11T21:52:44Z] it's going to take literal days [2026-03-11T21:53:18Z] what cpu did you say you have rn? [2026-03-11T21:54:11Z] I have built firefox on a lenovo X230. But honestly I don't remember how much time I needed. [2026-03-11T21:54:28Z] intel core ultra 5 125h. [2026-03-11T21:56:57Z] vouivre: https://www.cpubenchmark.net/compare/5544vs5777/nxp-IMX8MQ-vs-Intel-Ultra-5-125H [2026-03-11T21:57:22Z] my cpu is about 4% as capable as that one [2026-03-11T21:57:23Z] lol [2026-03-11T21:57:33Z] overall [2026-03-11T21:57:57Z] lol [2026-03-11T21:58:21Z] it's fine for small builds but i wouldn't do anything too crazy [2026-03-11T21:58:49Z] I was hoping maybe it can be done during the night [2026-03-11T21:58:53Z] no [2026-03-11T21:59:05Z] you'll have to cross-compile it or download a binary [2026-03-11T22:00:15Z] so a binary distro is the only choice [2026-03-11T22:00:22Z] also not true [2026-03-11T22:00:38Z] you just have to cross-compile the largest packages, like i said [2026-03-11T22:00:55Z] for most simple software it can be built on the device quite easily [2026-03-11T22:02:38Z] it's something new to me. But at home I could do it, I also have a desktop and I could do the cross-compilation. [2026-03-11T22:04:00Z] What do you use for the cross-compilation ? gcc ? [2026-03-11T22:04:47Z] yes [2026-03-11T22:04:56Z] you need to build a dedicated cross compiler though [2026-03-11T22:05:14Z] the script `mussel` automates building musl-based cross compilers [2026-03-11T22:06:13Z] ok [2026-03-11T22:06:48Z] thank you for all the information! [2026-03-11T22:07:00Z] I have to think about it [2026-03-11T22:33:36Z] np