mitsy

[awsdert]

Figured this be the right board for a project in C, anyways some people might have been interested in joining my hobby project they've seen me mention:
GitHub - awsdert/mitsy: MIT licensed C compiler
This thread is ONLY for mentioning your github username and any details that would identify your account from similar accounts, as a side note you'll need to have at least one open source repository if you're not a regular helper here since I will need to see you at least have enough experience programming to not make things worse on a regular basis.

[stahta01]

I think that me helping could hurt the project; since, I have done a few GCC debugging and applying other peoples patches I have seen GCC code base enough to make a accidental code copying a possible compliant. Tim S.

[awsdert]

I think that me helping could hurt the project; since, I have done a few GCC debugging and applying other peoples patches I have seen GCC code base enough to make a accidental code copying a possible compliant. Tim S.

Fair doos, well either way will be a while before I start encouraging others to join, this was mearly a chance for me to take notice of anyone that did want to join in, I'm aiming to get as far as the addon support stage before I really start to look for help since a compiler is too big a project to take on by oneself especially with my end goals at the moment

[stahta01]

Are you going to be using binutils or another similar library?
Edit: like The elfutils project
Or is it just going to target a single CPU family?

Tim S.

[stahta01]

musl - Wikipedia

musl is a C standard library intended for operating systems based on the Linux kernel, released under the MIT License.[3] It was developed by Rich Felker with the goal to write a clean, efficient and standards-conformant libc implementation.[4]

Just want to make sure you know about MUSL

[awsdert]

Are you going to be using binutils or another similar library?
Edit: like The elfutils project
Or is it just going to target a single CPU family?

Tim S.

I might use 'em, depends if they meat my needs, and no not gonna target just a single CPU, I'm starting with AMD64 cause that's what I got on my computer so I can test my code and resulting binaries before making more permanent changes, stability before portability is my motto here

musl - Wikipedia



Just want to make sure you know about MUSL

Didn't know about it, should be helpful I reckon

[awsdert]

musl - Wikipedia



Just want to make sure you know about MUSL

Now that I've actually looked at it (was on bus when made 1st answer) it's not greatly suitable for a compiler that aims to be able to compile kernels, in user space it is fine so I will start with that until I can build & run elf files from code (so that I know any bugs come from my code) but after that I will have to build in support for finding system libraries and redirecting to those when available, anyways thanks for the tip, I'll use it in other hobby projects anyways

[Hodor]

I've forgotten my github password and keys lmao

[awsdert]

I've forgotten my github password and keys lmao

How'd ya get in ya house without ya keys then?

[Hodor]

mcc_get.c line 272 (see also line 325 which I think exhibits the same problem)

if ( c >= 0xD000 || c <= 0xDFFF )

Surely this always evaluates to true. As an example, I'll change it to some different numbers so it's easier to type an example

if ( c >= 10 || c <= 100 )

Now, choose any random number and the expression will be true. c = 1, that's <= 100, so yep. c = 15, that's >= 10, so yep. c = 200, that's >= 10, so yep.

Maybe it's meant to be

if ( c >= 0xD000 && c <= 0xDFFF )

Edit:
I wrote a test program but gcc and clang don't spit out a warning (but all the numbers do result in the condition being true). Hmm.

#include <stdio.h>

int main(void)
{
    int i;
    
    for (i = 0; i < 1000; i++) {
        if (i >= 100 || i <= 200)
            printf("%3d Condition true\n", i);
    }
    
    return 0;
}

Edit 2: splint doesn't warn either. So maybe my logic is broken this morning (but I don't think so, and at least for the example program I wrote the condition is always true as I suspected)

[awsdert]

mcc_get.c line 272 (see also line 325 which I think exhibits the same problem)

if ( c >= 0xD000 || c <= 0xDFFF )

Surely this always evaluates to true. As an example, I'll change it to some different numbers so it's easier to type an example

if ( c >= 10 || c <= 100 )

Now, choose any random number and the expression will be true. c = 1, that's <= 100, so yep. c = 15, that's >= 10, so yep. c = 200, that's >= 10, so yep.

Maybe it's meant to be

if ( c >= 0xD000 && c <= 0xDFFF )

That's right thanks for pointing it out, didn't realise I slipped up there

[Hodor]

I see you applied the fix. You may have missed it, but I mentioned another line in the first post (kind of easy to miss consider I only mentioned it in parenthesis on the first line of the post hehe). It's now line 272

    if ( c >= 0xD000 || c <= 0xDFFF )          /* Same typo as previous typo (?); i.e. condition always true */
        return MCC_CHAR_TYPE_24BIT;

[Hodor]

Changes I made to get it to build

@@ -1,5 +1,6 @@
+#include <float.h>
 #include "mcc_get.h"
 int mcc_encoding_type = 0;
 int mcc_encoding_node = mcc_encoding_count;
 char const *env_LANG = NULL;
 mcc_char_info_t mcc_common_char_info[ mcc_encoding_count ] = {{0}};
@@ -442,14 +443,16 @@ int mcc_getall( MCC_POS *src, MCC_CH8 *dst,
     tok.src = mcc_iconv_tok_mem( tok.src, srcm->addr,
         srcv->use * mcc_char_info.size );
     if ( mcc_char_info.enc == mcc_encoding_ch8 ) {
         if ( srcv->use > (dstv->cap - dstv->use) ) {
             ret = mcc_vecsize( dstv,
-                (dstv->use + srcv->use) * sizeof(mcc_ch8_t),
+                (dstv->use + srcv->use + 1) * sizeof(mcc_ch8_t),
                 sizeof(mcc_ch8_t) );
             if ( ret != EXIT_SUCCESS ) return ret;
-        }
+        } else {
+            ret = EXIT_SUCCESS;
+        }
         tok.dst = mcc_iconv_tok_mem( tok.dst, dstm->addr, dstm->size );
         (void)memcpy( tok.dst.addr, srcm->addr, tok.src.size );
         dstv->use += srcv->use;
         goto mcc_getall_done;
     }
@@ -976,11 +979,11 @@ int mcc_getnum(
     long len = 0;
     ullong rev = 0, num = 0, dec, pos, bit, no_mul = 0;
     sllong exp;
     if ( !dst ) return EDESTADDRREQ;
     (void)memset( dst, 0, sizeof(MCC_NUM) );
-    if ( !src || !src->text.use ) return ENODATA;
+    if ( !src || !src->text.vec.use ) return ENODATA;
     if ( !base ) {
         if ( mcc_ch8ctype( C[0] ) == MCC_CHAR_TYPE_SPACE ) {
             mcc_getnum_sign:
             while ( (ret = mcc_getc(src, C, &len )) == EXIT_SUCCESS ) {
                 if ( len != 1 ) break;

The only dubious change is

(dstv->use + srcv->use + 1) * sizeof(mcc_ch8_t),

Which I merely added so that there's room for the '\0' character. The other changes were required to get it to actually build (apart from the ret = EXIT_SUCCESS; which is necessary to make sure it's initialised if the preceding condition is false)

[awsdert]

Changes I made to get it to build

@@ -1,5 +1,6 @@
+#include <float.h>
 #include "mcc_get.h"
 int mcc_encoding_type = 0;
 int mcc_encoding_node = mcc_encoding_count;
 char const *env_LANG = NULL;
 mcc_char_info_t mcc_common_char_info[ mcc_encoding_count ] = {{0}};
@@ -442,14 +443,16 @@ int mcc_getall( MCC_POS *src, MCC_CH8 *dst,
     tok.src = mcc_iconv_tok_mem( tok.src, srcm->addr,
         srcv->use * mcc_char_info.size );
     if ( mcc_char_info.enc == mcc_encoding_ch8 ) {
         if ( srcv->use > (dstv->cap - dstv->use) ) {
             ret = mcc_vecsize( dstv,
-                (dstv->use + srcv->use) * sizeof(mcc_ch8_t),
+                (dstv->use + srcv->use + 1) * sizeof(mcc_ch8_t),
                 sizeof(mcc_ch8_t) );
             if ( ret != EXIT_SUCCESS ) return ret;
-        }
+        } else {
+            ret = EXIT_SUCCESS;
+        }
         tok.dst = mcc_iconv_tok_mem( tok.dst, dstm->addr, dstm->size );
         (void)memcpy( tok.dst.addr, srcm->addr, tok.src.size );
         dstv->use += srcv->use;
         goto mcc_getall_done;
     }
@@ -976,11 +979,11 @@ int mcc_getnum(
     long len = 0;
     ullong rev = 0, num = 0, dec, pos, bit, no_mul = 0;
     sllong exp;
     if ( !dst ) return EDESTADDRREQ;
     (void)memset( dst, 0, sizeof(MCC_NUM) );
-    if ( !src || !src->text.use ) return ENODATA;
+    if ( !src || !src->text.vec.use ) return ENODATA;
     if ( !base ) {
         if ( mcc_ch8ctype( C[0] ) == MCC_CHAR_TYPE_SPACE ) {
             mcc_getnum_sign:
             while ( (ret = mcc_getc(src, C, &len )) == EXIT_SUCCESS ) {
                 if ( len != 1 ) break;

The only dubious change is

(dstv->use + srcv->use + 1) * sizeof(mcc_ch8_t),

Which I merely added so that there's room for the '\0' character. The other changes were required to get it to actually build (apart from the ret = EXIT_SUCCESS; which is necessary to make sure it's initialised if the preceding condition is false)

Could've sworn I already added that 1 elsewhere, whatever it's probably better to that anyways, also thanks for the point out on that typo with src->text.vec.use, finally the reason I didn't explicitly set ret was cause I was supposed to have filled it with mcc_char_info_test() return code and checked that first, thanks for indirectly pointing that out to me , Gotta post my intended thread this time.