Re: Hacking wwwstat..

Peter Seitz (seitz@bzs.tu-graz.ac.at)
Fri, 29 Aug 1997 13:06:25 +0200 

On Thu, 28 Aug 1997 12:26:09 -0700 (PDT)
   David Welton <davidw@efn.org> wrote:

> Hi again.. changes are progressing nicely on wwwstat in my spare time at
> work.  Recently we managed to add meaningful visits statistics to the
> archive sections - the commercial tool I have been comparing it against
> does not do this:-)
> 
> Having worked with the program for a while, it seems as if its big
> deficiency is data storage.  Reading its own output is definitely a 'cute
> perl trick', but, it is not very clean.  For example, there are currently
> || options so it can read the current output and the old output.  Do we
> keep doing this every time we change?  In addition, the regexps to read
> the data in are sort of yucky:P
> 
> At the suggestion of one of the other people here (Noah Couture - who also
> gets credit for the archive visits stuff), I started thinking about using
> dbm files, or maybe just doing some sort of tab delimited output.  This
> way the program could be broken into two pieces - one to process the log
> files, and one to generate output.  This would allow more, and easier
> manipulation of the results, into whatever for  you want - spreadsheets,
> databases, graphs, plain text like it is now, html tables.. whatever..
> 
> I have never really dealt with gdbm files, so I'm not really sure if maybe
> I'm barking up the wrong tree, and that the tabbed files are the best
> option.
> 
> What is the general opinion?

Well I am a starter with perl but I also like to extend some things 
in wwwstat. For example the internationalization is not fully 
implemented in wwwstat 2.0. I've found out, that there are still some 
strings in the main programm which do not allow a full output in a 
different langage other than english.

What I like to do is to supply a commandline switch so you can choose 
the language for the report at runtime. According to the language, 
a special config-file can be read in. But then I am sure that 
wwwstat will have troubles in reading older statistics, because the 
strings won't match.

So if you start creating a little database or a tab delimited output, 
this would be a big help for internationalized output of wwwstat (as 
you've already stated in your examples).

I would apreciate such an extension of wwwstat.

The other question is, what will then be the "official" version or 
let's say the "latest maintained" version of wwwstat. How can we 
ensure that there are not lots of people reinventing the wheel again, 
because they do not know that others already have implemented the 
feature they are working on. Maybe someone on the list knows how to 
coordinate the future development of wwwstat.


With best compliments
                       
           Peter Seitz
                                       (Faculty for civil engineering)
 -+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-  -+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-
  Verkehrswesenzeichensaal (VZS)   The better drawing site :-)

  University of Technology Graz, Technikerstr. 4/II, A-8010-Graz, Austria
  e-mail: mailto:seitz@bzs.tu-graz.ac.at (Internet)  2:316/5.18 (FidoNet)
 -+-+-+-+-+-+-+-+- http://wwwbzs.tu-graz.ac.at/~seitz/ +-+-+-+-+-+-+-+-+-