Thank for the insight.

Why C11 and not just C99 ?

Because C11 provided enough justifiable features and not being 27 years old.

Like name struct initializer.

I don't want to mess with C89.

The C "standard" is a joke that borrow "ideas" from other languages to try to 
stay "relevant".

Yet the definition of these features is vague at best, and left to the 
implementer interpretation at worse.

Like it mostly always did, hence the many incompatible compilers, ABI, etc.

That a source code base is "tied" to a C revision (your __thread example) shows 
how unreliable and weak is the C language.

I'm currently fighting with many datatype definitions in an Arduino program 
(char*, int8_t, TChar, BYTE, ...) because libraries uses different "standard".

I'm really considering to switch to something just as "old" but at least 
consistent : pascal.

FreePascal is an incredibly efficient compiler, and pascal code is a breeze to 
read and write.

Not like the usual mess of header inclusions and guards, macros targeting 
certain compiler version and vendor, etc.

Battling so much just to get a consistent "Hello World" out in 2026 gets me out 
of my mind.

All these issues should have been solved ages ago, not stacking up.

It's like it is becoming harder to write in C than in assembler.

And with that it's Friday the 13th.

No wonder the mood I'm in.

regards.




----- Mail d'origine -----
De: Michael Ackermann via Tinycc-devel <[email protected]>
À: [email protected]
Cc: Michael Ackermann <[email protected]>
Envoyé: Fri, 13 Feb 2026 13:33:05 +0100 (CET)
Objet: [Tinycc-devel] release tag criteria - Re:  Re : Re: Re :  The project 
seems to be missing a new release and/or a roadmap

Hi,

to summarize, again, i see two major criteria for a tinycc release tag made from
the system-integration perspective which covers many use-cases and implies
notable compile-time and run-time testing:

1. a complete system-integration of and with TinyCC for "boostrapping"
   which is documented among bootstrappable.org / live-bootstrap and related;
   recently bootstrappable-developers reported changes made to tinycc HEAD
   were helpful to avoid a dedicated tcc-fork for bootstrappable and they can
   roll along with mob HEAD since recently; nice.

2. spawning and maintaining a complete(!) TinyCC/Linux operating system
   distribution for at least one ARCH, that is x86_32 currently which i've
   backported and am maintaining more than 500 builds for already with some
   bootable kernel/libc and all development utilities available for self-hosting
   compilation; amazing.

Recently i've fixed several dozen bugs while runtime-testing the 
TinyCC/linux-2.4
system which was deployed into production as my secured main development host
during my last christmas hackathon. Keep in mind i backported most recent
userspace software versions back onto a 20year old kernel, in parallel to
swapping toolchain for tinycc in a complicated crossbuild setup, remaining
portable towards both linux5 and linux2 sysv ABI, portable towards different
ARCH=x86|arm|.. with different toolchains.
Many problems weren't related to tinycc, but linux2 ABI for example which is
unavoidable because i could not yet port any later kernel version with tcc.
sntpd/ntpclient didn't sync reliably, now it does, things like this which should
be ironed out before i586-tinycc-linux2-musl.iso got tagged for release.
Although the system is up and running with heavy compilation-load, heavy nfs/ssh
load, stable 24/7 without kernel panics, decent hardware support with some
drivers backported; but there is a few known nasty bugs/workarounds related to
tinycc remaining. I'm referring to a release-tag of a complete TinyCC/OS
distribution here, not tinycc itself which tinycc-developers should/must have
the last word spoken for then, not myself.

With regards to C-standardization followed, the current pragmatic approach had
proven appropriate to accomplish mentioned tasks above. Nonetheless C11 is
dangerous(!!!) if it destabilizes tinycc further than was necessary
(there's been some regression fixed recently which had cost me several nights
 http://tinyfront.mooo.com/docs.html#fnref:3)
Then, for example, i stumbled upon bind9 package, that i've compiled/linked
with tinycc, and finally picked some older version (9.11.37 2022) which did NOT
rely upon C11 __thread. A good example btw. how a complete system integration
could provide a baseline and confirmation for necessary C-language features.
C11 isn't a major criteria for c-toolchain portability of and with a complete
bootstrappable/C-portable GNUish/POSIX operating system distribution - not yet -
and that's good! Linux2 and NPTL sysv ABI compatibility is a whole other
discussion; i did rebase NPTL patchset for x86_32 onto tcc/linux-2.4,
and so far even NPTL tested OK with a 20year old kernel and two most recent
package-versions which need it (squashfs, tor-proxy). I too want to have a look
at the original tccboot again, two years ago i did rebase latest tcc HEAD with
it to compile linux-2.4 JiT (relevant to bootstrappable); it did work still but
i've had no time for this and focused on the AoT kbuild variant instead.

Anyway, among 500 builds / libc / kernel that i've already forked and patched
for tinycc-support, currently i'm hacking some old TeTeX-3.0 build to spawn
that one with tinycc too, for practical reasons if such a system was dedicated
to a LONG-TERM STABLE SCIENTIFIC DEVELOPMENT WORKSTATION. I am not yet aware of
any modern TeX system which could be "bootstrapped" and or remained portable
with a C-toolchain such as TinyCC. I'll too have a glimpse at QEMU again, to
compile/link that with tinycc, after all QEMU originated from Bellard just like
TinyCC, and it could be relevant to bootsrappable.org having QEMU
compiled/linked with tcc. Finally, if tinycc can be utilized as the main
system-compiler/toolchain for such a thing then it's probably good enough to
consider a release tag made, for ARCH=x86_32 at least.
mplayer/ffmpeg (too originated from Bellard) passed compile-time, but i've had
no time for runtime testing this either.

Long story short, kernel fork is uploaded already for regular compile-time and
runtime testing, and i too want to and will publish/release the userspace fork
once a few bureaucratic/political issues got coped with in germany which
sabotaged my work for almost 20years for political bargain here. The userspace
fork is zipped into a fully(!) forked gentoo-portage tree regardless for
comfortable tracking and cryptographically secured build-log archival. Nowadays
it's just not feasible anymore to merge for tinycc/c-toolchain portability
upstream with them, GNU/gcc and llvm/clang and with it free/open software are
heading into a direction i won't follow along. Just saying. Hence the
tinycc-driven system is fully self-hosting
(and too "bootstrappable"! outside the project scope of mine) including all
required development utilities (gdb,strace,python,perl,autotools) since 
recently.

Given this, in advance of a complete TinyCC/OS released, i want to express my
gratitude towards tinycc-developers who made this possible.

Documentation is online with a tiny test site at
http://tinyfront.mooo.com/docs.html

Friendly Greetings

Michael Ackermann

On 2026-02-13 09:04, [email protected] wrote:
> Tcc 'mob' has became a project of many goals.
> 
> Some like it the way it is because it suits their purpose (K&R C would do, 
> C89 too).
> 
> Some try to use it for RISC-V compilation to avoid toolchains of pachydermic 
> dimension (LLVM, GCC).
> 
> I wanted to have full C11 compliance because I dislike tcc hybrid-like nature 
> of the feature set and standard support.
> 
> Let's says I "hoped" for tcc not to fall in the same demise than minix, but 
> that's not my main quest in the end.
> 
> Some people in the list told me that tcc is quite beloved around, despite not 
> having a clear versioning scheme.
> 
> I too wanted a "clarification" of tcc status, showing it being maintained, 
> kept up to date and "relevant".
> 
> In the end it's grishka, the main maintainer in charge, who decide if a new 
> release is worth the pain.
> 
> There are other similar compilers out there after all, even I use VBCC 
> instead for my m68k developments :
> 
> http://www.compilers.de/vbcc.html
> 
> There are 8 billions people out there, I think if each one did his part, the 
> world wouldn't be in the state it is.
> 
> Same for open source projects.
> 
> Regards.
> 
> 
> ----- Mail d'origine -----
> De: Eyal Rozenberg <[email protected]>
> À: david koch <[email protected]>, [email protected]
> Envoyé: Fri, 13 Feb 2026 01:10:04 +0100 (CET)
> Objet: Re: Re : [Tinycc-devel] The project seems to be missing a new release 
> and/or a roadmap
> 
> Indeed, I was not aware of the history of past discussions. However - no 
> user who does not 'mine' the mailing list is aware of that. So, My 
> request and suggestion remains...
> 
> At the link, there is a thread about the criteria for releasing version 
> 1.0, and it is said that "fully passing conformance tests" is a 
> necessary condition for that. I see that there was some discussion on 
> alternative goals for a 1.0 version, e.g. being able to faithfully 
> compile a working version of a suite of programs. Did you end up 
> reaching a consensus? Also, even if you weren't - there is not even a 
> proper table of missing/implemented/partial values for features by C 
> standard. Here is GCC's:
> 
> https://gcc.gnu.org/projects/c-status.html
> 
> there are a lot more C99 features that are mentioned in the 
> documentation for example. Why not duplicate this page, start by filling 
> in what's already documented (either with the first implementing release 
> or starting with "sometime before 0.9.27" to be made exact later), and 
> proceed to the other features? And put that on the tinycc.org website, 
> with links from savanna, the repo website(s), or even the README?
> 
> 
> You suggested that I define a clear goal for the next version. Well, I'm 
> not a TinyCC developer, so I don't feel it is my place to do that. But - 
> here is a possible goal from a user's perspective: Endow the user base 
> with the improvements to the compiler between 2017 and some point near 
> the present - in the form of a release, which is of a stability/quality 
> comparable or better than that of 0.9.27. Developers are able to just 
> pull the latest head of the main branch and use it reasonably and 
> carefully, but most (?) users typically limit themselves to releases, as 
> head commits sometimes contain glaring ephemeral issues introduced 
> recently by mistake, or new features that are semi-broken. And that is 
> obviously much less likely when you create a release branch and freeze 
> it for new features, landing only bug fixes on it, for a while.
> 
> 
> And even a decision about a roadmap or upcoming minor release is not 
> made soon - it is still important to update the website with where 
> things stand: Describe how the project governance works now (as opposed 
> to the self-disciplined free push policy), mention some key active 
> contributors perhaps. I see some of this in:
> 
> https://www.mail-archive.com/[email protected]/msg11284.html
> 
> and of course, the README still has Fabrice Bellard signing as what seem 
> the single maintainer/developer.
> 
> Respectfully,
> Eyal
> 
> On 09/02/2026 16:08, [email protected] wrote:
> > Hello you,
> > 
> > this discussion **again** ?
> > 
> > https://www.mail-archive.com/[email protected]/msg11276.html
> > 
> > Define a clear goal for the next version and do your part to make tcc reach 
> > it.
> > 
> > If you followed list closely, you'd seen there are still many bugs.
> > 
> > Yes, I'm old and grumpy.
> > 
> > David
> > 
> > 
> > ----- Mail d'origine -----
> > De: Eyal Rozenberg via Tinycc-devel <[email protected]>
> > À: TinyCC Development Mailing List <[email protected]>
> > Cc: Eyal Rozenberg <[email protected]>
> > Envoyé: Mon, 09 Feb 2026 14:50:40 +0100 (CET)
> > Objet: [Tinycc-devel] The project seems to be missing a new release and/or 
> > a roadmap
> > 
> > Hello all,
> > 
> > I am not a tinycc contributor (and only rarely a user). I joined the
> > mailing list to relay a sentiment of the "casual user".
> > 
> > So, I'm looking at the project's:
> > 
> > Website       https://tinycc.org
> > repo-website  https://repo.or.cz/w/tinycc.git
> > README
> > 
> > and here is what I notice:
> > 
> > * The last release, 0.9.27, was made 9 years ago, in 2017. Nearly a
> > decade. This, while development and maintenance seems to be continuing
> > unabated :-)
> > * The website has not been updated since 2018.
> > * None of the three venues give an up-to-date description of the project
> > in the  structural/organizational sense, i.e. how are decisions and and
> > by whom.
> > * None of the three venues contain a roadmap, outlook or plan for future
> > activities, specifically a new release.
> > 
> > So, I would like to ask that you consider:
> > 
> > * Preparing/committing a prospective roadmap regarding future work and
> > release schedule for the project.
> > * Making a 0.9.28 or 0.10.0 (or 1.0 ?) release, soon.
> > * Updating all three venues with some or or most of that information.
> > 
> > 
> > Respectfully,
> > 
> > Eyal Rozenbergx     


_______________________________________________
Tinycc-devel mailing list
[email protected]
https://lists.nongnu.org/mailman/listinfo/tinycc-devel

Reply via email to