Guys, tinycc in it's recent form and shape already is capable of compiling and statically linking a complete distribution including all required development utilities. There's a few limitations with regards to "bootstrapping", which i would consider highly relevant too, as a system-integration path for TinyCC itself which bootstrappable.org maintained. Yet i fear tinycc might become de-stabilized with new features added, before some known-good working baseline was fully established with a complete distribution. Some regressions were noticed already. And bootstrappable.org had to maintain a tcc fork too iirc.
The whole approach is documented at the test-site. I merely hesitate with the devdrop upload, because i've begun intense runtime testing, by utilizing the mentioned devdrop as a buildhost itself, seeing to _all_ development tooling remained self-hosting and stable... most of which are, such as python, perl, autotools, gdb, strace, ... except bash.static which blocks against bashism inside portage tooling that is needed to handle the complexity of toolchain portability and support for various ARCH, compatibility with either linux2 and linux5 ABI, and full chrooted cross-compiling support etc. etc. Next, i want to avoid confusion and wasting time of other developers, with problems i rather cope with and fix myself before. Furthermore, it's a few GiB of source distfiles and ISOs, that aren't easily provided with a 2Mbps uplink and dynamic IP address, besides time and effort to keep things in order with release management. And finally, if i'm hit by another round of extortion letters and threats from bureaucracy and politics in germany, i'll have to worry about other problems. That's no bad intention on my side at all, the bootable devdrop wasn't uploaded yet. You've no idea what type of bureaucratic nonsense i was molested with, for month and years. Finally, since i've had no better things to do, i had to bake bread in grocery store in the morning, and hack tinyfront OS at night. In return for being told all that's worth nothing and less than minimum wage. A large stack of inquiries sent to several dozens of employers and three universities yielded no response at all, none, germany. In the meantime, until and if at all things got settled, i think the linux-tcc kernel build could be used for regular compile-time and run-time testing already, hence doesn't need much time and effort for regular CI test iterations. For the sake of it this kernel even got some cheapo usb2.0 ethernet dongle support back on board, which too was a little time and distraction. The recommended approach would be, to fully stabilize and keeping the known-good working baseline intact, and keeping _the_ working kernel while new (optional) features get introduced for C11/linux5. Because, many years ago a guy named seyko2 (who contributed to tcc too), already had established a linux-2.4 fork, but it wasn't maintained and finally broke again with latest tcc. That wasn't trivial to bisect and debug, for month. Otherwise i see no problem with improving TinyCC, if working things don't break. I git-pulled tinycc sources myself once or twice a year only, because i've had to rebase a few patches for the build-system and tried to keep the fully forked portage tree in sync with main gentoo tree, although a full fork and various blockers weren't avoidable anymore. Its a huge fork, and merging back with gentoo is not possible. All of this takes time and money which i haven't got, nor the stamina to explain to gentoo USE=-cxx with tcc was high priority. And I'm not in the mood to discuss at LKML. Of cause userspace parts including the static musl libc.a shouldn't break either, however the portage tree fork isn't ready for git-push, it's barely good enough for the regular devdrops i compiled. To answer the question: bugs and features to cope with are collected inside the portage tree fork with hundreds of ebuilds and patches, plus the documentation that was uploaded. In the near future I hope tinycc devs may utilize the distribution for easy re-production of bugs (such as the bash.static one currently which i can't at least summarize a reasonable error report for yet, because it's totally erratic) I'll keep you updated once this issue got resolved. Regards. On 2025-08-09 22:52, Robin Rowe wrote: > On 8/9/2025 4:18 PM, Michael Ackermann via Tinycc-devel wrote: > > I would propose a trivial criteria for the required C language feature set: > > a _complete_ operating system distribution must compile and become bootable > > as > > first priority with whatever tinycc was capable to compile and link already. > > Good idea. List of bugs/features that tcc needs addressed to achieve > this goal? > > Robin > > _______________________________________________ > Tinycc-devel mailing list > Tinycc-devel@nongnu.org > https://lists.nongnu.org/mailman/listinfo/tinycc-devel
signature.asc
Description: Digital signature
_______________________________________________ Tinycc-devel mailing list Tinycc-devel@nongnu.org https://lists.nongnu.org/mailman/listinfo/tinycc-devel
