Bug? Intermittent situation with CGI program

andu undo at cloud9.net
Mon Jul 15 19:23:00 EDT 2002


--On Monday, July 15, 2002 13:54:28 -0700 Sadhunathan Nadesan 
<sadhu at castandcrew.com> wrote:

> Aloha,
>
> We have identified an intermittent bug situation using MC as a CGI
> language which I will submit to the bug list.  Personally I don't read
> the bug list emails and I imagine there are others who don't either
> but who might like to know of the existence of the situation.
>
> In a nutshell it is that using MC as a CGI to collect data from a form
> (which is in turn dynamically generated), it sometimes fails to
> collect anything, on some large forms.  It is as if the form on the
> web page did a transmit, but, the MC CGI did not do a receive.
>
> I realize that itermittent bugs are very hard to track down.  We can
> provide a test situation wherein, in one of our offices (NY) the cgi
> always fails on a given form, for certain users running Outlook 2002
> or Outlook 2000, and Windows 98, sometimes fails for other users in
> this office (same set up), and never fails for other users in our San
> Diego office with the same environment.
>
> This would seem to suggest the environment!  However, when we recode
> the same CGI program in Perl, it always works for all users.  HMMMM!!
> This might suggest something coded wrong or lacking in the MC program.
>
> In other words, it might not really be a bug in MC.  I'm sending it in
> anyway.
>

<snip>

Two things I noticed in the last mail which may or may not have anything to 
do with your problem: when the script "puts" the reply I usually give 
"Content_Length=length_of_reply & crlf" also and use crlf for the header 
part instead of cr, the other thing

put "</html>" & cr
put "</body>" & cr

should actually be reversed though I don't think this is the problem.
Also might help to read from stdin until eof or empty rather then for 
$CONTENT_LENGTH. Also note that the shellCommand defaults to bash so you 
don't have to set it all the time, just do shell().

In any case I doubt there is a bug with MC, most likely the script 
somewhere.

Regards, Andu Novac



More information about the metacard mailing list