Forum OpenACS Q&A: What happened to servername-error.log

we're running open acs 3/aolserver 3.2 and doing heavy tcl/sql
development of a number of new modules using adp pages, and tcl
routines.
we're doing the usual - window 1 browser, window 2, tail -f
servername.log.  The issue is that the sql errors get swallowed up in
the noise and its tough to find -
We had a classic ACS 3.5/Oracle install that had 2 logs -
servername.log and servername-error.log - what ever happened to that?
AND as long as we're on this topic - theres gotta be a better way to
debug code.
Collapse
Posted by C. R. Oldham on
<em>
theres gotta be a better way to debug code.
</em>
<p>
Yes--in another thread today someone was using TclPro to debug, but the code needed some massaging before you could break into it with TclPro.  That really excites me--is there an (easy) way we could make that work with the templating system?
</p>
Collapse
Posted by Jerry Asher on
I never got TclPro and AOLserver to work 100% of the time.  It seemed that every other breakpoint was caught, but I suspect that was due to operator error.
<p>
Yes there has got to be a better way to debug code.  When developing pages, I feel that I've fallen back 20 years to
<pre>
10 PRINT $I
</pre>
I have a bit more sophisticated version of that.  The combination of <tt>nscp</tt>, emacs, and David Lutterkort's emacs server log module, <a href='http://www.arsdigita.com/acs-repository/one-archive?archive_id=441';>http://www.arsdigita.com/acs-repository/one-archive?archive_id=441</a>, are a win for me.
<p>
I typically have one emacs window on the code, an emacs window shelled into AOLserver itself through <tt>nscp</tt>, an emacs window serving the log, and an emacs window in <tt>SQL</tt> mode.
<p>
Truly the big win is <tt>nscp</tt> mode which lets me evaluate arbitrary AOLserver tcl expressions and communicate with the running AOLserver itself.  It's a Lisp Interaction buffer supported by AOLserver.
<p>
I also find that having browsers open to the Tcl manual at activestate, the aolserver manual at aolserver, and a Guide to Semiconductor Physics at britneyspears.ac is also helpful.
Collapse
Posted by Mat Kovach on
Actually I've spend some much time in the "older languages" (FORTRAN, COBOL) that I'm usually using:

WRITE(9,100)

WRITE(9,900) (CURVEXY/PI), (CURVEZY/ 2 * CURVRVEZX / PI)

WRITE(9,101)

100 FORMAT('DEBUG: START')

101 FORMAT('DEBUG: STOP')

900 FORMAT('DEBUG: INPUT 1: ' F19.8 ' INPUT 2: ' E10.4)

So, I'm quite use to it.  Sometimes it takes wrapping things with easily grepped syntax.

tail -f error.log | grep 'STRING' works fairly good to clear the noise.

Collapse
Posted by S. Y. on
Jerry, you forgot to mention the Cisco documentation at: http://www.routergod.com