Re: `program_invocation_name' bug =?=> `gnomelib_init' API change

[sopwith redhat com (Elliot Lee)]

On 22 Nov 1998 00:02:26 -0600, Raja R Harinath <harinath@cs.umn.edu> wrote:
>Hi,
>
>I was experiencing a lot of SIGSEGVs recently.  While trying to figure
>this out I came across one bug that could be causing some. 
>
>On non-glibc systems, the variable `program_invocation_name' is not set
>until `gnomelib_parse_args' is called.  By this time `gnome_client_init'
>is already called.  `gnome_client_init' hence initializes
>`master_client->program' to a NULL value, which when used unwarily (as
>in passing to printf or strrchr), can cause a SIGSEGV.
>
>One hacky way to solve this is in `gnome_master_client' to verify if
>`master_client->program' is NULL, and re-initialize it there.  
>
>A better way, IMHO, is to pass just argv[0] to `gnomelib_init'.  This
>however involves changing the `gnomelib_init' API.  AFAI can make out,
>this is a low impact change, since I could find only 10 direct calls to
>`gnomelib_init', 5 of which were in the test-suite.
>
>Comments, alternative?

The first parameter to 'gnomelib_init' should be the app ID ("gnumeric",
"test-gnome", whatever) instead of being the basename(executable_name). So it
probably needs to be a string constant...

-- Elliot
"In film you will find four basic story lines. Man versus man, man
 versus nature, nature versus nature, and dog versus vampire."
    - Steven Spielberg