Re: [Tinycc-devel] Using tinycc for full source bootstrapping
On Wed, Sep 27, 2017 at 10:34:15AM +0200, Kamil Rytarowski wrote: > On 27.09.2017 06:55, Christian Jullien wrote: > > 1) fully supported on Linux on x86/x64/arm/aarch64 > > 2) fully supported on Windows x86/x64 > > 3) Unclear or not fully supported on *BSD and macOS, My different attempts > > on FreeBSD/OpenBSD/NetBSD all failed from one reason or another > > For NetBSD we need compiler builtin for va_list: ... > Without this we cannot use TCC with our native headers. Generally, the native headers do not have to be used. As a workaround it is possible to use tcc on NetBSD via LinuxABI with a C library like musl, with clean, not kernel-specific includes. Even natively on NetBSD we could still use tcc, as long as the actual kernel keeps backward compatibility with the version which we target. We would have to use our own custom headers expressing the correct ABI in a suitable fashion, like musl does on Linux. This may be hard but possibly not very hard, Rune ___ Tinycc-devel mailing list Tinycc-devel@nongnu.org https://lists.nongnu.org/mailman/listinfo/tinycc-devel
Re: [Tinycc-devel] Using tinycc for full source bootstrapping
On 27.09.2017 06:55, Christian Jullien wrote: > I have the impression that this discussion is more philosophic than > pragmatic. > > Today tcc compiles and runs on limited number of systems/processors which > are AFAIK: > > 1) fully supported on Linux on x86/x64/arm/aarch64 > 2) fully supported on Windows x86/x64 > 3) Unclear or not fully supported on *BSD and macOS, My different attempts > on FreeBSD/OpenBSD/NetBSD all failed from one reason or another > For NetBSD we need compiler builtin for va_list: #ifdef __lint__ typedef char *__va_list; #else typedef __builtin_va_list __va_list; #endif -- sys/ansi.h Without this we cannot use TCC with our native headers. signature.asc Description: OpenPGP digital signature ___ Tinycc-devel mailing list Tinycc-devel@nongnu.org https://lists.nongnu.org/mailman/listinfo/tinycc-devel
Re: [Tinycc-devel] Using tinycc for full source bootstrapping
I have the impression that this discussion is more philosophic than pragmatic. Today tcc compiles and runs on limited number of systems/processors which are AFAIK: 1) fully supported on Linux on x86/x64/arm/aarch64 2) fully supported on Windows x86/x64 3) Unclear or not fully supported on *BSD and macOS, My different attempts on FreeBSD/OpenBSD/NetBSD all failed from one reason or another On source code I see also C67 but I have seen no discussions on it for years. That said, for 1, 3, standard C compiler that comes with OS supports C98 for nearly 20 years. For 2, VC++, clang, gcc, mingw32, Cygwin support also C98 for nearly 20 years. As tcc developer time and resources are limited, except for the fun, who can tell me a **real in use system** that needs to bootstrap tcc with a C compiler that only accepts C89? Of course, when compiled, I fully agree that tcc should accept C89 to support legacy code that no one wants to change. If one finds, *still in production*, a very exotic system that supports only C89, the solution is to cross compile tcc on, says Linux, then use this version to bootstrap tcc natively. For years, gcc has the nice feature to make Canadian cross-compilation that I once use to generate a working gcc on Solaris sparc from Linux x86. So my conclusion is that I see no reasons to limit ourselves to C89. -Original Message- From: Tinycc-devel [mailto:tinycc-devel-bounces+eligis=orange...@nongnu.org] On Behalf Of KHMan Sent: mercredi 27 septembre 2017 04:38 To: tinycc-devel@nongnu.org Subject: Re: [Tinycc-devel] Using tinycc for full source bootstrapping On 9/27/2017 10:26 AM, Michael B. Smith wrote: > As I read your comments, especially these: > >>> I would find it awesome if tinycc would restrict its source to C89. >> >> 2017 - 1989 = 28. >> >> Twenty-Eight Years. It would be kinda nice to move forward. Will we stick to C89 simply because of certain platforms that may never move forward? It's kinda like tying stones to one's legs. Or a lowest common denominator scenario. > > You were suggesting to abandon the "old" or "legacy" support. Context. The discussion was on tcc bootstrapping. The question I was posing was whether tcc should be tied to a commitment to a 28 year old standard. English can be a very ambiguous language. A tiny thought can lead to tunnel vision down one of many supposedly plausible paths. > If I misunderstood, please forgive me. > > -Original Message- > From: Tinycc-devel On Behalf Of KHMan > Sent: Tuesday, September 26, 2017 10:18 PM > To: tinycc-devel@nongnu.org > Subject: Re: [Tinycc-devel] Using tinycc for full source bootstrapping > [snipped everything] -- Cheers, Kein-Hong Man (esq.) Selangor, Malaysia ___ Tinycc-devel mailing list Tinycc-devel@nongnu.org https://lists.nongnu.org/mailman/listinfo/tinycc-devel ___ Tinycc-devel mailing list Tinycc-devel@nongnu.org https://lists.nongnu.org/mailman/listinfo/tinycc-devel
Re: [Tinycc-devel] Using tinycc for full source bootstrapping
On 9/27/2017 10:26 AM, Michael B. Smith wrote: As I read your comments, especially these: I would find it awesome if tinycc would restrict its source to C89. 2017 - 1989 = 28. Twenty-Eight Years. It would be kinda nice to move forward. Will we stick to C89 simply because of certain platforms that may never move forward? It's kinda like tying stones to one's legs. Or a lowest common denominator scenario. You were suggesting to abandon the "old" or "legacy" support. Context. The discussion was on tcc bootstrapping. The question I was posing was whether tcc should be tied to a commitment to a 28 year old standard. English can be a very ambiguous language. A tiny thought can lead to tunnel vision down one of many supposedly plausible paths. If I misunderstood, please forgive me. -Original Message- From: Tinycc-devel On Behalf Of KHMan Sent: Tuesday, September 26, 2017 10:18 PM To: tinycc-devel@nongnu.org Subject: Re: [Tinycc-devel] Using tinycc for full source bootstrapping [snipped everything] -- Cheers, Kein-Hong Man (esq.) Selangor, Malaysia ___ Tinycc-devel mailing list Tinycc-devel@nongnu.org https://lists.nongnu.org/mailman/listinfo/tinycc-devel
Re: [Tinycc-devel] Using tinycc for full source bootstrapping
As I read your comments, especially these: >> I would find it awesome if tinycc would restrict its source to C89. > > 2017 - 1989 = 28. > > Twenty-Eight Years. It would be kinda nice to move forward. Will we stick to > C89 simply because of certain platforms that may never move forward? It's > kinda like tying stones to one's legs. Or a lowest common denominator > scenario. You were suggesting to abandon the "old" or "legacy" support. If I misunderstood, please forgive me. -Original Message- From: Tinycc-devel [mailto:tinycc-devel-bounces+michael=theessentialexchange@nongnu.org] On Behalf Of KHMan Sent: Tuesday, September 26, 2017 10:18 PM To: tinycc-devel@nongnu.org Subject: Re: [Tinycc-devel] Using tinycc for full source bootstrapping On 9/27/2017 9:54 AM, Michael B. Smith wrote: > You funny. > > I've still got hundreds of thousands of lines of C in K&R C. In production. > Supporting major applications. > > You are obviously one of those people who thinks that COBOL is ancient and > unused, aren't you? > > Because I've got millions of production LOC in COBOL. > > No offense, but I don't think you understand the real world. Please look at this line from Rune: "Compilers written in C89 and understanding C99 hardly exist." This implies that his intention is to get C99 via a C89 route. If his ending point is a C99 capable tcc and natively hosted, then it has nothing to do with any amount of legacy code, which can continue to use legacy compilers. I say certain platforms may never move forward because in those cases we are usually at the mercy of proprietary compilers and the companies that make them. We are not talking about a C89 end point here. C89 folks continue to use their stuff, I'm sure it will be in use past 2100. No problem with that. tcc was already using // comments with nary an outcry for like ages. When was it ever a problem? The only problem this time is that someone wants to bootstrap without any tiny bit of any kind of cross compiling variations. No offense, but I don't think I have been out of line. > -Original Message- > From: Tinycc-devel On Behalf Of KHMan > Sent: Tuesday, September 26, 2017 9:41 PM > To: tinycc-devel@nongnu.org > Subject: Re: [Tinycc-devel] Using tinycc for full source bootstrapping > > On 9/27/2017 5:30 AM, u-j...@aetey.se wrote: >> On Sun, Sep 24, 2017 at 07:40:32PM +0200, grischka wrote: >>> Also as source tcc is supposed to be C89, except maybe 'long long' >>> and maybe some minor things here or there which we could change >>> indeed if that is wanted (for example usage of compound initializers >>> in arm-gen.c) >> >> [not a tinycc developer but using / depending on compiler >> bootstrapping] >> >> I would find it awesome if tinycc would restrict its source to C89. > > 2017 - 1989 = 28. > > Twenty-Eight Years. It would be kinda nice to move forward. Will we stick to > C89 simply because of certain platforms that may never move forward? It's > kinda like tying stones to one's legs. Or a lowest common denominator > scenario. > >> Compilers written in C89 and understanding C99 hardly exist. This >> makes it hard to do a proper bootstrap from the ground with minimal >> dependency on existing binaries. > > Could you give an example/examples where the starting point for a > bootstrap is C89 for the foreseeable future? (Not trying to be a > nitpick pest, but actual data points are usually a good thing, and tcc > supports only a few processors, so it would be nice to know which one > and what platform, or is it some future thing.) > > I like Larry's posting. Item 1 can also be done by text processing. > > Or make a minimally functioning tcc C89 branch. > > Or bootstrap it the Pascal P4 way. > >> (Offtopic, but tightly related to the full bootstrapping: if tcc >> would also become relicensed to BSD-alike, this would not only make >> such deep bootstrapping easier but also allow getting to C99 without >> relying on GPL, which matters to some people and scenarios) >> >> Thanks for your work on tcc, >> Rune -- Cheers, Kein-Hong Man (esq.) Selangor, Malaysia ___ Tinycc-devel mailing list Tinycc-devel@nongnu.org https://lists.nongnu.org/mailman/listinfo/tinycc-devel ___ Tinycc-devel mailing list Tinycc-devel@nongnu.org https://lists.nongnu.org/mailman/listinfo/tinycc-devel
Re: [Tinycc-devel] Using tinycc for full source bootstrapping
On 9/27/2017 9:54 AM, Michael B. Smith wrote: You funny. I've still got hundreds of thousands of lines of C in K&R C. In production. Supporting major applications. You are obviously one of those people who thinks that COBOL is ancient and unused, aren't you? Because I've got millions of production LOC in COBOL. No offense, but I don't think you understand the real world. Please look at this line from Rune: "Compilers written in C89 and understanding C99 hardly exist." This implies that his intention is to get C99 via a C89 route. If his ending point is a C99 capable tcc and natively hosted, then it has nothing to do with any amount of legacy code, which can continue to use legacy compilers. I say certain platforms may never move forward because in those cases we are usually at the mercy of proprietary compilers and the companies that make them. We are not talking about a C89 end point here. C89 folks continue to use their stuff, I'm sure it will be in use past 2100. No problem with that. tcc was already using // comments with nary an outcry for like ages. When was it ever a problem? The only problem this time is that someone wants to bootstrap without any tiny bit of any kind of cross compiling variations. No offense, but I don't think I have been out of line. -Original Message- From: Tinycc-devel On Behalf Of KHMan Sent: Tuesday, September 26, 2017 9:41 PM To: tinycc-devel@nongnu.org Subject: Re: [Tinycc-devel] Using tinycc for full source bootstrapping On 9/27/2017 5:30 AM, u-j...@aetey.se wrote: On Sun, Sep 24, 2017 at 07:40:32PM +0200, grischka wrote: Also as source tcc is supposed to be C89, except maybe 'long long' and maybe some minor things here or there which we could change indeed if that is wanted (for example usage of compound initializers in arm-gen.c) [not a tinycc developer but using / depending on compiler bootstrapping] I would find it awesome if tinycc would restrict its source to C89. 2017 - 1989 = 28. Twenty-Eight Years. It would be kinda nice to move forward. Will we stick to C89 simply because of certain platforms that may never move forward? It's kinda like tying stones to one's legs. Or a lowest common denominator scenario. Compilers written in C89 and understanding C99 hardly exist. This makes it hard to do a proper bootstrap from the ground with minimal dependency on existing binaries. Could you give an example/examples where the starting point for a bootstrap is C89 for the foreseeable future? (Not trying to be a nitpick pest, but actual data points are usually a good thing, and tcc supports only a few processors, so it would be nice to know which one and what platform, or is it some future thing.) I like Larry's posting. Item 1 can also be done by text processing. Or make a minimally functioning tcc C89 branch. Or bootstrap it the Pascal P4 way. (Offtopic, but tightly related to the full bootstrapping: if tcc would also become relicensed to BSD-alike, this would not only make such deep bootstrapping easier but also allow getting to C99 without relying on GPL, which matters to some people and scenarios) Thanks for your work on tcc, Rune -- Cheers, Kein-Hong Man (esq.) Selangor, Malaysia ___ Tinycc-devel mailing list Tinycc-devel@nongnu.org https://lists.nongnu.org/mailman/listinfo/tinycc-devel
Re: [Tinycc-devel] Using tinycc for full source bootstrapping
You funny. I've still got hundreds of thousands of lines of C in K&R C. In production. Supporting major applications. You are obviously one of those people who thinks that COBOL is ancient and unused, aren't you? Because I've got millions of production LOC in COBOL. No offense, but I don't think you understand the real world. -Original Message- From: Tinycc-devel [mailto:tinycc-devel-bounces+michael=theessentialexchange@nongnu.org] On Behalf Of KHMan Sent: Tuesday, September 26, 2017 9:41 PM To: tinycc-devel@nongnu.org Subject: Re: [Tinycc-devel] Using tinycc for full source bootstrapping On 9/27/2017 5:30 AM, u-j...@aetey.se wrote: > On Sun, Sep 24, 2017 at 07:40:32PM +0200, grischka wrote: >> Also as source tcc is supposed to be C89, except maybe 'long long' >> and maybe some minor things here or there which we could change >> indeed if that is wanted (for example usage of compound initializers >> in arm-gen.c) > > [not a tinycc developer but using / depending on compiler > bootstrapping] > > I would find it awesome if tinycc would restrict its source to C89. 2017 - 1989 = 28. Twenty-Eight Years. It would be kinda nice to move forward. Will we stick to C89 simply because of certain platforms that may never move forward? It's kinda like tying stones to one's legs. Or a lowest common denominator scenario. > Compilers written in C89 and understanding C99 hardly exist. This > makes it hard to do a proper bootstrap from the ground with minimal > dependency on existing binaries. Could you give an example/examples where the starting point for a bootstrap is C89 for the foreseeable future? (Not trying to be a nitpick pest, but actual data points are usually a good thing, and tcc supports only a few processors, so it would be nice to know which one and what platform, or is it some future thing.) I like Larry's posting. Item 1 can also be done by text processing. Or make a minimally functioning tcc C89 branch. Or bootstrap it the Pascal P4 way. > (Offtopic, but tightly related to the full bootstrapping: if tcc would > also become relicensed to BSD-alike, this would not only make such > deep bootstrapping easier but also allow getting to C99 without > relying on GPL, which matters to some people and scenarios) > > Thanks for your work on tcc, > Rune -- Cheers, Kein-Hong Man (esq.) Selangor, Malaysia ___ Tinycc-devel mailing list Tinycc-devel@nongnu.org https://lists.nongnu.org/mailman/listinfo/tinycc-devel ___ Tinycc-devel mailing list Tinycc-devel@nongnu.org https://lists.nongnu.org/mailman/listinfo/tinycc-devel
Re: [Tinycc-devel] Using tinycc for full source bootstrapping
On 9/27/2017 5:30 AM, u-j...@aetey.se wrote: On Sun, Sep 24, 2017 at 07:40:32PM +0200, grischka wrote: Also as source tcc is supposed to be C89, except maybe 'long long' and maybe some minor things here or there which we could change indeed if that is wanted (for example usage of compound initializers in arm-gen.c) [not a tinycc developer but using / depending on compiler bootstrapping] I would find it awesome if tinycc would restrict its source to C89. 2017 - 1989 = 28. Twenty-Eight Years. It would be kinda nice to move forward. Will we stick to C89 simply because of certain platforms that may never move forward? It's kinda like tying stones to one's legs. Or a lowest common denominator scenario. Compilers written in C89 and understanding C99 hardly exist. This makes it hard to do a proper bootstrap from the ground with minimal dependency on existing binaries. Could you give an example/examples where the starting point for a bootstrap is C89 for the foreseeable future? (Not trying to be a nitpick pest, but actual data points are usually a good thing, and tcc supports only a few processors, so it would be nice to know which one and what platform, or is it some future thing.) I like Larry's posting. Item 1 can also be done by text processing. Or make a minimally functioning tcc C89 branch. Or bootstrap it the Pascal P4 way. (Offtopic, but tightly related to the full bootstrapping: if tcc would also become relicensed to BSD-alike, this would not only make such deep bootstrapping easier but also allow getting to C99 without relying on GPL, which matters to some people and scenarios) Thanks for your work on tcc, Rune -- Cheers, Kein-Hong Man (esq.) Selangor, Malaysia ___ Tinycc-devel mailing list Tinycc-devel@nongnu.org https://lists.nongnu.org/mailman/listinfo/tinycc-devel
Re: [Tinycc-devel] Using tinycc for full source bootstrapping
On Sun, Sep 24, 2017 at 07:40:32PM +0200, grischka wrote: > Also as source tcc is supposed to be C89, except maybe 'long long' > and maybe some minor things here or there which we could change > indeed if that is wanted (for example usage of compound initializers > in arm-gen.c) [not a tinycc developer but using / depending on compiler bootstrapping] I would find it awesome if tinycc would restrict its source to C89. Compilers written in C89 and understanding C99 hardly exist. This makes it hard to do a proper bootstrap from the ground with minimal dependency on existing binaries. (Offtopic, but tightly related to the full bootstrapping: if tcc would also become relicensed to BSD-alike, this would not only make such deep bootstrapping easier but also allow getting to C99 without relying on GPL, which matters to some people and scenarios) Thanks for your work on tcc, Rune ___ Tinycc-devel mailing list Tinycc-devel@nongnu.org https://lists.nongnu.org/mailman/listinfo/tinycc-devel
Re: [Tinycc-devel] Using tinycc for full source bootstrapping
I second grischka, Unless you prove it is buggy, the best thing to do with obscure code you don't understand IS TO KEEP IT AS IS! I apply this advice to my own code :o)) Doing otherwise is the fasted way to introduce new bugs that no one will find until several months. -Original Message- From: Tinycc-devel [mailto:tinycc-devel-bounces+eligis=orange...@nongnu.org] On Behalf Of grischka Sent: mardi 26 septembre 2017 09:19 To: tinycc-devel@nongnu.org Subject: Re: [Tinycc-devel] Using tinycc for full source bootstrapping Larry Doolittle wrote: > The comma operator example you posted, that confused me at first, is > an exception, and I just pushed a patch. You override someone else's style decision with your own because you are confused? We're all confused, eventually. -- gr ___ Tinycc-devel mailing list Tinycc-devel@nongnu.org https://lists.nongnu.org/mailman/listinfo/tinycc-devel ___ Tinycc-devel mailing list Tinycc-devel@nongnu.org https://lists.nongnu.org/mailman/listinfo/tinycc-devel
Re: [Tinycc-devel] Using tinycc for full source bootstrapping
Larry Doolittle wrote: The comma operator example you posted, that confused me at first, is an exception, and I just pushed a patch. You override someone else's style decision with your own because you are confused? We're all confused, eventually. -- gr ___ Tinycc-devel mailing list Tinycc-devel@nongnu.org https://lists.nongnu.org/mailman/listinfo/tinycc-devel
Re: [Tinycc-devel] Using tinycc for full source bootstrapping
On Sun, Sep 24, 2017 at 07:40:32PM +0200, grischka wrote: > Jan Nieuwenhuizen wrote: > >Are you willing to consider discussing/taking patches that reduce the > >variety and complexity of C constructs used in TinyCC source code? > >I have a list of about 20 unpolished patches[0] for constructs like: > I wonder which would be the smaller patch: The changes you propose > for tcc or a patch to support these constructs as is in the bootstrap > compiler. I wonder that too. I looked through a lot of the changes you made, and 99% of them decrease the code quality of tcc. IMHO the long-term path forward would be to improve your bootstrap compiler. The comma operator example you posted, that confused me at first, is an exception, and I just pushed a patch. Using semicolons here is more of a "say what you mean" and "principle of least surprise" approach to coding this function. The _other_ comma operators you found (and eliminated) in tcc's code base are used for all the right reasons, to reduce code duplication and create a smooth flow of control. I enthusiastically support boostrap projects in general, and wish you the best of luck going forward! Just don't expect mainline tcc to regress its code quality for your benefit. - Larry ___ Tinycc-devel mailing list Tinycc-devel@nongnu.org https://lists.nongnu.org/mailman/listinfo/tinycc-devel
Re: [Tinycc-devel] Using tinycc for full source bootstrapping
Friends - On Sun, Sep 24, 2017 at 12:27:17PM -0700, Larry Doolittle wrote: > On Sun, Sep 24, 2017 at 09:48:25AM +0200, Jan Nieuwenhuizen wrote: > > -p[0] = x & 255, p[1] = x >> 8 & 255; > The original line above, however, screams "I am a bug" to me. > The return value of "x & 255" is ignored, and it has no side effects, > so it can be discarded, leaving >p[0] = p[1] = x >> 8 & 255; > Is there any evidence it works? I tested it, and it does in fact work. Sorry, I got a little mixed up on the order of operations. Still, using comma there instead of semicolon seems needlessly confusing. It suggests that these were once tricky macro expansions instead of inline functions. If it were me I would use semicolon there (write32le()) and in write64le(). - Larry ___ Tinycc-devel mailing list Tinycc-devel@nongnu.org https://lists.nongnu.org/mailman/listinfo/tinycc-devel
Re: [Tinycc-devel] Using tinycc for full source bootstrapping
Friends - On Sun, Sep 24, 2017 at 09:48:25AM +0200, Jan Nieuwenhuizen wrote: > -p[0] = x & 255, p[1] = x >> 8 & 255; > +p[0] = x & 255; p[1] = x >> 8 & 255; I think the comma operator is an awesome part of the C standard, and sometimes permits much more clear and concise expression of process than would otherwise be possible. The original line above, however, screams "I am a bug" to me. The return value of "x & 255" is ignored, and it has no side effects, so it can be discarded, leaving p[0] = p[1] = x >> 8 & 255; Is there any evidence it works? - Larry ___ Tinycc-devel mailing list Tinycc-devel@nongnu.org https://lists.nongnu.org/mailman/listinfo/tinycc-devel
Re: [Tinycc-devel] Using tinycc for full source bootstrapping
I wonder which would be the smaller patch: The changes you propose for tcc or a patch to support these constructs as is in the bootstrap compiler. Otherwise if that is not your criterion what would be the arguments for the semantics that you mention that they do not belong to "simple enough C"? Of course from our point of view tcc IS a simple C compiler almost by definition, so, since it can compile itself, it must be written in already sufficiently simple C ;) Also as source tcc is supposed to be C89, except maybe 'long long' and maybe some minor things here or there which we could change indeed if that is wanted (for example usage of compound initializers in arm-gen.c) -- gr Jan Nieuwenhuizen wrote: Hi! Are you willing to consider discussing/taking patches that reduce the variety and complexity of C constructs used in TinyCC source code? I have a list of about 20 unpolished patches[0] for constructs like: comma operator -p[0] = x & 255, p[1] = x >> 8 & 255; +p[0] = x & 255; p[1] = x >> 8 & 255; or (heterogeneous) initialzer lists -uint8_t buf[1000], *p = buf; +uint8_t buf[1000]; +uint8_t *p = buf; or function return derefencing -v = tok_alloc(astr.data, astr.size - 1)->tok; +TokenSym *tk = tok_alloc(astr.data, astr.size - 1); +v = tk->tok; or complex field access -return table_ident[v]->sym_define; +TokenSym *t = table_ident[v]; +Sym *s = t->sym_define; +return s; or even struct by value assign -vtop->type = *type; +memcpy (&vtop->type, type, sizeof (CType)); that allow tcc to be compiled with a simpler C compiler, so that in time we could add to README something like - BOOTSTRAPPABLE! Because tcc uses only a subset of C99 (C99--), it can be bootstrapped with a simple C compiler. Knowing[1] that TinyCC can build Gcc (wow!) and building tcc is easier than building gcc, it is a good candidate for use in a bootstrap path. The past couple of months I have been working to bootstrap tcc using mescc, a simple C compiler that can (almost) be bootstrapped from source. I have just released[2] Mes 0.10 which can compile a modified version of mes-tcc, which can compile a trivial C program into a working executable. Much work will have to be done before mes-tcc will be able to build a functional Gcc. I would like to start by reducing the delta on tcc that I'm carrying. What do you think, is bootstrappable builds[4] something you would want to support and what could C99-- look like? Greetings, janneke [0] https://gitlab.com/janneke/tinycc/commits/wip-mescc [1] https://lists.gnu.org/archive/html/tinycc-devel/2017-05/msg00099.html [2] http://lists.gnu.org/archive/html/guile-user/2017-09/msg00027.html [3] http://bootstrappable.org ___ Tinycc-devel mailing list Tinycc-devel@nongnu.org https://lists.nongnu.org/mailman/listinfo/tinycc-devel
Re: [Tinycc-devel] Using tinycc for full source bootstrapping
A non-maintainer, here are my 2c: As a matter of style, I like this idea but: - tcc is probably already the smallest C compiler that compiles a decent C compiler - as you probably noticed, there is not a huge activity on tcc these days and I personally prefer than one spent time on something immediately useful for tcc core, like C11 support. - changing code that works is risky because we must ensure nothing is broken and tcc tests suite is far from complete (not counting the fact that we have no platform regression tests) - it's a common practice for a compiler X to be compiled by itself. This has the advantage to test more code combinations it is supposed to support. - it will introduce gratuitous diff that will make harder to identify real regression That said, and especially because the lack of a decent non-regression tests suite, I personally vote against your proposal. -Original Message- From: Tinycc-devel [mailto:tinycc-devel-bounces+eligis=orange...@nongnu.org] On Behalf Of Jan Nieuwenhuizen Sent: dimanche 24 septembre 2017 09:48 To: tinycc-devel@nongnu.org Cc: Jeremiah Orians Subject: [Tinycc-devel] Using tinycc for full source bootstrapping Hi! Are you willing to consider discussing/taking patches that reduce the variety and complexity of C constructs used in TinyCC source code? I have a list of about 20 unpolished patches[0] for constructs like: comma operator -p[0] = x & 255, p[1] = x >> 8 & 255; +p[0] = x & 255; p[1] = x >> 8 & 255; or (heterogeneous) initialzer lists -uint8_t buf[1000], *p = buf; +uint8_t buf[1000]; +uint8_t *p = buf; or function return derefencing -v = tok_alloc(astr.data, astr.size - 1)->tok; +TokenSym *tk = tok_alloc(astr.data, astr.size - 1); +v = tk->tok; or complex field access -return table_ident[v]->sym_define; +TokenSym *t = table_ident[v]; +Sym *s = t->sym_define; +return s; or even struct by value assign -vtop->type = *type; +memcpy (&vtop->type, type, sizeof (CType)); that allow tcc to be compiled with a simpler C compiler, so that in time we could add to README something like - BOOTSTRAPPABLE! Because tcc uses only a subset of C99 (C99--), it can be bootstrapped with a simple C compiler. Knowing[1] that TinyCC can build Gcc (wow!) and building tcc is easier than building gcc, it is a good candidate for use in a bootstrap path. The past couple of months I have been working to bootstrap tcc using mescc, a simple C compiler that can (almost) be bootstrapped from source. I have just released[2] Mes 0.10 which can compile a modified version of mes-tcc, which can compile a trivial C program into a working executable. Much work will have to be done before mes-tcc will be able to build a functional Gcc. I would like to start by reducing the delta on tcc that I'm carrying. What do you think, is bootstrappable builds[4] something you would want to support and what could C99-- look like? Greetings, janneke [0] https://gitlab.com/janneke/tinycc/commits/wip-mescc [1] https://lists.gnu.org/archive/html/tinycc-devel/2017-05/msg00099.html [2] http://lists.gnu.org/archive/html/guile-user/2017-09/msg00027.html [3] http://bootstrappable.org -- Jan Nieuwenhuizen | GNU LilyPond http://lilypond.org Freelance IT http://JoyofSource.com | Avatar® http://AvatarAcademy.com ___ Tinycc-devel mailing list Tinycc-devel@nongnu.org https://lists.nongnu.org/mailman/listinfo/tinycc-devel ___ Tinycc-devel mailing list Tinycc-devel@nongnu.org https://lists.nongnu.org/mailman/listinfo/tinycc-devel
Re: [Tinycc-devel] Using tinycc for full source bootstrapping
I'm not an active contributor so my opinion doesn't count for much, but here it is anyway. Though I like the idea of implementing TCC in a smaller subset of C it should be a clearly defined subset of C and preferably one that makes sense generally rather than just for a particular bootstrap path. Why not start with C89 rather than C99? Some of the examples you gave don't make the language significantly simpler, in my opinion. I don't think restricting C's expression syntax makes the compiler writer's job any easier (though some code style rules reject the comma operator anyway). If you want to simplify things ban floating-point types, structs as arguments and most of the standard header files and library functions. Also "switch" and "goto", perhaps. Edmund ___ Tinycc-devel mailing list Tinycc-devel@nongnu.org https://lists.nongnu.org/mailman/listinfo/tinycc-devel
[Tinycc-devel] Using tinycc for full source bootstrapping
Hi! Are you willing to consider discussing/taking patches that reduce the variety and complexity of C constructs used in TinyCC source code? I have a list of about 20 unpolished patches[0] for constructs like: comma operator -p[0] = x & 255, p[1] = x >> 8 & 255; +p[0] = x & 255; p[1] = x >> 8 & 255; or (heterogeneous) initialzer lists -uint8_t buf[1000], *p = buf; +uint8_t buf[1000]; +uint8_t *p = buf; or function return derefencing -v = tok_alloc(astr.data, astr.size - 1)->tok; +TokenSym *tk = tok_alloc(astr.data, astr.size - 1); +v = tk->tok; or complex field access -return table_ident[v]->sym_define; +TokenSym *t = table_ident[v]; +Sym *s = t->sym_define; +return s; or even struct by value assign -vtop->type = *type; +memcpy (&vtop->type, type, sizeof (CType)); that allow tcc to be compiled with a simpler C compiler, so that in time we could add to README something like - BOOTSTRAPPABLE! Because tcc uses only a subset of C99 (C99--), it can be bootstrapped with a simple C compiler. Knowing[1] that TinyCC can build Gcc (wow!) and building tcc is easier than building gcc, it is a good candidate for use in a bootstrap path. The past couple of months I have been working to bootstrap tcc using mescc, a simple C compiler that can (almost) be bootstrapped from source. I have just released[2] Mes 0.10 which can compile a modified version of mes-tcc, which can compile a trivial C program into a working executable. Much work will have to be done before mes-tcc will be able to build a functional Gcc. I would like to start by reducing the delta on tcc that I'm carrying. What do you think, is bootstrappable builds[4] something you would want to support and what could C99-- look like? Greetings, janneke [0] https://gitlab.com/janneke/tinycc/commits/wip-mescc [1] https://lists.gnu.org/archive/html/tinycc-devel/2017-05/msg00099.html [2] http://lists.gnu.org/archive/html/guile-user/2017-09/msg00027.html [3] http://bootstrappable.org -- Jan Nieuwenhuizen | GNU LilyPond http://lilypond.org Freelance IT http://JoyofSource.com | Avatar® http://AvatarAcademy.com ___ Tinycc-devel mailing list Tinycc-devel@nongnu.org https://lists.nongnu.org/mailman/listinfo/tinycc-devel
