On Mon, Oct 29, 2018 at 4:46 PM Denys Vlasenko <vda.li...@googlemail.com> wrote:
>
> On Mon, Oct 29, 2018 at 4:55 PM Gavin Howard <gavin.d.how...@gmail.com> wrote:
> >
> > Hello,
> >
> > Because my patch is so large, I cannot get it through the mailing list
> > filter. Therefore, I have pasted it on pastebin, instead.
> >
> > The link is https://pastebin.com/PJa2vaUR and the raw version can be
> > found at https://pastebin.com/raw/PJa2vaUR
>
> Probably could send it bzipped.
I tried, unfortunately. Even though it was under the limit, the
mailing list software rejected.
> All functions except bc_main() should be static.
> Check with nm --size-sort bc.o
I will see what I can do about that.
> Use <libbb.h>: the entire
>
> +BcStatus bc_read_file(const char *path, char **buf) {
> +
> + BcStatus s = BC_STATUS_IO_ERR;
> + FILE *f;
> + size_t size, read;
> + long res;
> + struct stat pstat;
> +
> + if (!(f = fopen(path, "r"))) return BC_STATUS_EXEC_FILE_ERR;
> +
> + if (fstat(fileno(f), &pstat) == -1) goto malloc_err;
> +
> + if (S_ISDIR(pstat.st_mode)) {
> + s = BC_STATUS_PATH_IS_DIR;
> + goto malloc_err;
> + }
> +
> + if (fseek(f, 0, SEEK_END) == -1) goto malloc_err;
> + if ((res = ftell(f)) < 0) goto malloc_err;
> + if (fseek(f, 0, SEEK_SET) == -1) goto malloc_err;
> +
> + if (!(*buf = malloc((size = (size_t) res) + 1))) {
> + s = BC_STATUS_ALLOC_ERR;
> + goto malloc_err;
> + }
> +
> + if ((read = fread(*buf, 1, size, f)) != size) goto read_err;
>
> can be replaced by single xmalloc_open_read_close()!
Will do.
> (The entire
> +#include <ctype.h>
> +#include <errno.h>
> +#include <stdbool.h>
> +#include <stddef.h>
> +#include <stdint.h>
> +#include <stdio.h>
> +#include <stdlib.h>
> +#include <string.h>
> +
> +#include <fcntl.h>
> +#include <limits.h>
> +#include <signal.h>
> +#include <unistd.h>
> +#include <sys/stat.h>
> +#include <sys/types.h>
> +
> +#include <getopt.h>
> should be replaced by
> #include "libbb.h"
> #include <if_you_need_anything_not_already_included_by_libbb.h>
> )
Sorry about that. I tried to follow the documentation as exactly as I could.
> + while ((c = getopt_long(argc, argv, bc_args_opt, bc_args_lopt,
> &idx)) != -1)
> + {
>
> Use getopt32() ?
Okay. Will do.
> +typedef struct BcResult {
> +
> + BcResultType t;
> + BcResultData d;
> +
> +} BcResult;
>
> Lets drop these extra empty lines.
Will do.
> +#ifdef CONFIG_BC
> +BcStatus bc_vm_posixError(BcStatus s, const char *file,
> + size_t line, const char *msg);
> +#endif // CONFIG_BC
>
> "ifdef CONFIG_foo" is unsafe versus typos. Use "if ENABLE_foo" - it
> gives warnings
> for typos.
Oh? Did not know that.
> +#ifdef CONFIG_BC
> +
> +#endif // CONFIG_BC
>
> Why these empty ifdef blocks?
This was caused by bugs in my release script that have been fixed.
> +const char bc_name[] = "bc";
>
> static
>
>
> +const char *bc_errs[] = {
>
> static const char *const bc_errs[] = {
Will do.
> + if (!(v->v = malloc(esize * BC_VEC_START_CAP))) return
> BC_STATUS_ALLOC_ERR;
>
> (1) Please do not nest assignments:
>
> v->v = malloc(esize * BC_VEC_START_CAP);
> if (!v->v)
> return BC_STATUS_ALLOC_ERR;
Oh boy. I was afraid of this.
The problem is that I am also trying to get this bc into toybox (might
as well reduce duplication of effort), and as I am sure you know,
Landley is very particular about the line count of commands. To be
honest, I would prefer your style (except for putting if statements
that on one line if possible), but this reduces line count.
I don't really know what to do here.
> (2) Use xmalloc/xzalloc/xrealloc which never return NULL.
> We do not try to survive out-of-memory. No check is necessary.
Will do. I was not aware of those functions. bc does have to print a
diagnostic message (see
http://pubs.opengroup.org/onlinepubs/9699919799/utilities/bc.html#tag_20_09_15),
but I checked that those procedures do that.
> +BcStatus bc_vec_pushByte(BcVec *v, uint8_t data) {
> + return bc_vec_push(v, &data);
> +}
>
> The entire bbox code base uses this formatting of function body:
>
> BcStatus bc_vec_pushByte(BcVec *v, uint8_t data)
> {
> return bc_vec_push(v, &data);
> }
This one might be a little harder.
> + if (!v->len && (s = bc_vec_pushByte(v, '\0'))) return s;
>
> len is not boolean or a pointer. !foo is usually used with those.
> len is an integer. You are testing whether it's zero. Why obfuscate?
> For readability, should be:
> if (v->len == 0) {
> s = bc_vec_pushByte(v, '\0');
> if (s)
> return s;
> }
>
>
> + if (a == b) return 0;
> + else if (!a->len) return BC_NUM_NEG(!!b->len, !b->neg);
> + else if (!b->len) return BC_NUM_NEG(1, a->neg);
> + else if (a->neg) {
> + if (b->neg) neg = true;
> + else return -1;
> + }
> + else if (b->neg) return 1;
>
> "else" after exiting code in preceding "if" is not necessary.
> Can rewrite as:
>
> if (a == b) return 0;
> if (!a->len) return BC_NUM_NEG(!!b->len, !b->neg);
> if (!b->len) return BC_NUM_NEG(1, a->neg);
> if (a->neg) {
> if (!b->neg) return -1;
> neg = true;
> }
> else if (b->neg) return 1;
You are correct.
> +static void bc_vm_sig(int sig) {
> + if (sig == SIGINT) {
> + size_t len = strlen(bcg.sig_msg);
> + if (write(2, bcg.sig_msg, len) == (ssize_t) len)
> + bcg.sig += (bcg.signe = bcg.sig == bcg.sigc);
> + }
> + else bcg.sig_other = 1;
> +}
>
> + sa.sa_handler = bc_vm_sig;
> + sa.sa_flags = 0;
> +
> + if (sigaction(SIGINT, &sa, NULL) < 0 || sigaction(SIGPIPE, &sa,
> NULL) < 0 ||
> + sigaction(SIGHUP, &sa, NULL) < 0 || sigaction(SIGTERM, &sa, NULL) <
> 0)
> + {
> + return BC_STATUS_EXEC_SIGACTION_FAIL;
> + }
>
> Calculator has signal handling?!
> siganction() never fails - no need to check result.
GNU bc has signal handling. It is so users can stop long-running
calculations. This bc is meant to be compatible.
> Better yet, use read_line_input() from libbb.h:
> /*
> * maxsize must be >= 2.
> * Returns:
> * -1 on read errors or EOF, or on bare Ctrl-D,
> * 0 on ctrl-C (the line entered is still returned in 'command'),
> * >0 length of input string, including terminating '\n'
> */
> int read_line_input(line_input_t *st, const char *prompt, char
> *command, int maxsize) FAST_FUNC;
>
> - it already has Ctrl-C handling, and drop entire signal handling mess.
> (It's buggy: write() can modify errno).
Oh, thanks for the heads up about write. I will take care of that, but
to remain compatible with GNU, I will still have to do a write.
> Let bc just die on SIGPIPE, SIGTERM and SIGHUP.
I guess I could do that.
Okay, as it stands, I see one big showstopper for getting this into
busybox: code style. I don't know how to fix code style with a script,
so if it is a showstopper for you (and if it is, I don't blame you), I
guess I have to respectfully withdraw my patch and apologize for
wasting your time.
If, however, it is not, we can talk about what to do about it. And I
will get started on changing the other things.
Gavin Howard
_______________________________________________
busybox mailing list
busybox@busybox.net
http://lists.busybox.net/mailman/listinfo/busybox
