Date: Mon, 18 May 1998 18:52:33 +0300 (IDT)
From: Eli Zaretskii <eliz AT is DOT elta DOT co DOT il>
To: Pierre Muller <muller AT janus DOT u-strasbg DOT fr>
cc: djgpp AT delorie DOT com
Subject: Re: utod with LFN=Y bug ?
In-Reply-To: <199805181535.RAA29821@isis.u-strasbg.fr>
Message-ID: <Pine.SUN.3.91.980518184040.19060B-100000@is>
MIME-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Precedence: bulk


On Mon, 18 May 1998, Pierre Muller wrote:

>    And there I discovered a bug :
>    sizeof(try_char) returned 36
>    but this is not correct
>     because
>     static char try_char[] = "abcdefghijklmnopqrstuvwxyz012345789";
>     is only 35 bytes long (notice the missing 6 in the digits !!)

No, this is perfectly correct: sizeof includes the terminating null 
character, that's why the code says "sizeof(try_char)-1".  (The 
omission of 6 is probably a typo, but it shouldn't do any real harm in 
this case.)

>     so the statement 
> 
>     do
>     {
>       if (idx <= 0)
> 	     return -1;
>       *pbase = try_char[--idx];
>       /* bug sizeof(try_char) returns 36 instead of 35)
>       so the first is 0 !! */

No.  idx is initialized like this:

	int idx = sizeof(try_char)-1;

Since sizeof returns 36, idx begins with the value of 35.  So the 
statement

       *pbase = try_char[--idx];

actually does this:

       *pbase = try_char[34];

(since --idx means "decrement first, and then use").  So there's no 
problem here that I can see.

>    Morality : we should not use sizeof here
> because it round up to 4 bytes but strlen I suppose

No, sizeof doesn't round up, it just gives the size of the object in 
bytes.  sizeof can only round up when the compiler adds padding to a 
variable due to alignment considerations, but these do not apply to char 
buffers.

>    Adding the extra check worked for me
>    replacing sizeof by strlen worked also
>    reinserting the 6 digit would also work ... 

I cannot believe that this solved the problem.  I also don't see the
problem; I have run several different cases, just to be sure, and couldn't
get utod or dtou to fail.  Please post the details as I asked for them
(below).  You have left them without any response.  The most important
piece of information is the exact comaand line that you use when these
programs fail. 

> >There's something strange in your system setup, or in the files that you 
> >are trying to convert.  Please post the exact command you used to convert 
> >the files, and also the shortest file that can be used to reproduce your 
> >problem, including its file name.
> >
> >Please also post the output of "dir dtou.exe".