Re: [Refdb-users] User feedback 3

[David N. <dav...@bi...>]

Markus Hoenicka wrote:

>Thanks for that. I've added refdbxmlrc with some cosmetic changes to
>CVS. I've also integrated your refdbxml changes into the official
>version which also required some Makefile.am and configure.in
>hacking. I've fixed a few obvious typos in the file you sent and made
>some minor changes, like parametrizing the xep root variable. I've
>tested the updated script with xsltproc and FOP. I'm currently
>updating my xerces and Saxon installations, but I expect this will
>work too.
>  
>
I'm afraid that further testing has shown up some problems with my 
refdbxml script.  While I thoroughly tested the pdf output, I didn't 
test rtf output.  I'm doing so now and have discovered some errors:

1. The test for valid fo_processor does not include 'jfor'.  Here are 
the problem lines:

.........................................................................................................
if [ ! $fo_processor = "passivetex" ] && [ ! $fo_processor = "fop" ] && 
[ ! $fo_processor = "xep" ]; then
  echo "specify one of 'passivetex', 'fop', 'xep' with the -f option"
.........................................................................................................

and here's how it should read:

.........................................................................................................................................
if [ ! $fo_processor = "passivetex" ] && [ ! $fo_processor = "fop" ] && 
[ ! $fo_processor = "xep" ] && [ ! $fo_processor = "jfor" ]; then
                                                                                                   
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  echo "specify one of 'passivetex', 'fop', 'xep', 'jfor' with the -f 
option"
                                                 ^^^^^^^^
.........................................................................................................................................


2. The test for whether classpaths were extracted from init file is 
defective.  The problem lines are here:

.............................................................
# Set classpath variables to default if not already specified
if ! [ -n "$xslt_processor" ]; then
                 ^^^^^^^^^
    xslt_processor=$classpath
         ^^^^^^^^^
fi
if ! [ -n "$fo_processor" ]; then
               ^^^^^^^^^
    fo_processor=$classpath
       ^^^^^^^^^
fi
.............................................................

and the corrected lines here:

.............................................................
# Set classpath variables to default if not already specified
if ! [ -n "$xslt_classpath" ]; then
                 ^^^^^^^^^
    xslt_classpath=$classpath
         ^^^^^^^^^
fi
if ! [ -n "$fo_classpath" ]; then
               ^^^^^^^^^
    fo_classpath=$classpath
       ^^^^^^^^^
fi
.............................................................


3. The jfor command still used the (old) default classpath instead of 
the fo classpath.  The offending lines:

...............................................................................
process_print () {
    case $fo_processor in
    jfor )
        java -cp "$classpath" ch.codeconsult.jfor.main.CmdLineConverter 
$1 $2;;
...............................................................................

and the corrected version:

..................................................................................
process_print () {
    case $fo_processor in
    jfor )
        java -cp "$fo_classpath" 
ch.codeconsult.jfor.main.CmdLineConverter $1 $2;;
                   ^^^
..................................................................................


You may have picked up some (or all) of these when you diplomatically 
refer to "obvious typos".

Apologies for the inadequate testing.

By the way, the current jfor is little difficult to get running, mainly 
owing to inadequate (actually, mostly absent) documentation.  I found 
the clue I needed in a jfor user-group posting (archived on GMane) about 
the Windows version!  It's obviously nothing to do with refdb directly, 
except that the Makefile relies on jfor for rtf output.  I'll post 
details direct to refdb-users with an informative heading for anyone 
else who's trying to set jfor up.

Regards,
David.