Linux CGI blog.r

[jfdutcher]

I run blog.r without pain on Windows based web servers. To date no Linux based server will run blog.r successfully... after copying (local server) or FTPing (remote server) the same blog.r script to its CGI folder/directory where Rebol core exists.

All return External Server Error 500......no logs exist as Rebol is unsupported on the hoster officially.

What is the best thing I can do to effectively debug/capture the offending code ?

It can be noted that simple scripts (even ones that read/write files) do execute on the Linux servers so my environment must be O.K. I presume blog.r does some things that these simpler scripts do not....that the servers find offensive.

[PeterWood]

John

You could take a look at Sunanda's CGI-debug script at http://www.rebol.org/cgi-bin/cgiwrap/rebol/view-script.r?script=cgi-debug.r

It may help you to find the error.

Peter

[Sunanda]

Other things to try:

Does the script have the right line endings? It may not, if uploaded from WIN to LIN. The wrong line terminations mean your shebang line will  not be recognised properly.

Does the script use some modern/recent REBOL syntax that is not recognised by the version you are using on the server? If so, REBOL will fail when loading the script. The only example I know of is the improved path notation:
   a: 1 b: [1 2 3 4]
   b/:a: 9999   ;; variable in a path used in an assignment

Does the script have a needs header that is later than the version you run on the server?

[jfdutcher]

I have uploaded cgi-debug.r and included do %cgi-debug.r in several scripts on the server. All of them run as before......except the problematic one, blog.r.

It reports the same "Internal Server Error" message 500, as it did without the cgi-debug.r. Again I run the blog.r containing cgi-debug.r on the local "Windows" based server and it runs as before without incident.

The Rebol core module on the remote host running Linux was downloaded within the month. I'm not currently familiar with the 'needs' header....but I'll pursue that
issue....the 'header' is really whatever Carl had in the header when I downloaded blog.r a couple or three months ago.

[jfdutcher]

I just thought other potential users of the Apache Web Server on Linux or elsewhere would find the following 'real' eplanation for my problem of interest:


Apache mod_security uses rules to determine if a web request is "dangerous", and how to deal with such requests.
Bad rules will think that posts which contain "cat", "sh" followed by a space, "system", or any of thousands
of other possible word combinations are bad, and simply prevent your script from seeing the request, which causes
Apache to not get headers back from the script (it never runs), and generate an Internal Server Error 500.

GoDaddy level 1 support people have no real knowledge and are probably doing ther only thing they know how to do....state that the problem is a scripting error, which, of course is not the case.