bug#52438: 29.0.50; [PATCH] Fix off-by-one error in etags.c TeX support

[David Fussner]

Hi Lars,

> I'm not able to reproduce the issue on my 64-bit Debian system, but
> reading the code, I think your analysis of the problem sounds correct.
> The patch seemed slightly confusing, though, so I rewrote it a bit.
> Could you check whether this version also fixes the problem?

If my code was merely confusing and inelegant rather than flat-out
incorrect then that's a triumph :-)

> diff --git a/lib-src/etags.c b/lib-src/etags.c
> index bd4d4fcf53..ba66eeede4 100644
> --- a/lib-src/etags.c
> +++ b/lib-src/etags.c
> @@ -5773,7 +5773,7 @@ TeX_commands (FILE *inf)
>  TEX_decode_env (const char *evarname, const char *defenv)
>  {
>    const char *env, *p;
> -  ptrdiff_t len;
> +  ptrdiff_t len = 1;
>
>    /* Append default string to environment. */
>    env = getenv (evarname);
> @@ -5782,8 +5782,13 @@ TEX_decode_env (const char *evarname, const char 
> *defenv)
>    else
>      env = concat (env, defenv, "");
>
> +  /* If the environment variable starts with a colon, increase the
> +     length of the token table.  */
> +  if (!strneq (env, ":", 1))
> +    len++;
> +
>    /* Allocate a token table */
> -  for (len = 1, p = env; (p = strchr (p, ':')); )
> +  for (p = env; (p = strchr (p, ':')); )
>      if (*++p)
>        len++;
>    TEX_toktab = xnew (len, linebuffer);
>

That fixes the issue here, thank you very much! ( I guess Andreas'
comment on my strneq test is well-founded, though?)

Please close this bug if/when you apply your fix.

David.