To the ftp site manager - sorry, but I'm a bit ignorant of the correct
procedures for submitting things at your site.  Hope this is sufficient
information.

The thing I've uploaded is a metric tool based on Terence Parr's PCCTS.
The right file is cccc.tgz.2 (cccc.tgz was an abortive attempt which
I made without switching out of ASCII mode, and of course I can't 
overwrite it).  The file is a gzip'ed tar file (made on Linux with GNU tar).
below is a draft of the email I sent to Terence Parr about the file.  Please 
get in touch with me if you need more info, or if you wish to reject the file
(can't think why you would, but it's your site...)

thanks for your help


Tim Littlefair

t.littlefair@cowan.edu.au
tim@mitswa.com.au
83 The Strand
Bedford  WA 6052
Western Australia

---------


Terence

Just a quick note to thank you for PCCTS (which I've been lurking away
working on for most of this year), and to describe some of the product which
I've uploaded to ftp.parr-research.com for your consideration.

The stuff is in the file cccc.tgz and purports to be a metric analyser for
C++ code.  It implements a very coarse parser, which is designed (??) to
break an input file up into regions, each of which contains a single C++ 
class, method or instance variable definitions or declaration.  The program
counts lines of code, lines of comment, and McCabe's cyclomatic number for
each region it encounters (the counting rules are lexically based, and are 
pragmatic rather than exact). I don't think it's the greatest thing since
sliced bread, but the grammar may be of some interest to people with similar
coarse parsing requirements (reverse engineering applications spring to
mind), as it is smaller and probably more easy to understand than a full
grammar for the language needs to be, and it is designed (there's that word
again, but I only remember coding) to run on code which has not been
preprocessed, so it will recognize a method definition as such from the
pattern A::B without needing to have previously been shown a definition of A
as a class.

I have implemented the analyser as part of my M.Sc. thesis project, and
propose to release it for free general use on the internet in the near future
(i.e. after my supervisor has had a week or so to review the document
'project.txt' in the distribution, which is the written documentation we are
using to launch the first phase of the project).  My intentions in making
the tool available are not purely altruistic - I hope to be able to recruit 
subject sites for the later phases of my research from users of the package.

I am uploading it to you in advance to ask your advice on the most appropriate
site to list in the documentation as the primary source.  All my development
has been on Linux, and I certainly plan uploading a set of the final
distribution to sunsite.unc.edu.  I have got a copy of the latest LSM (Linux
System Map - i.e. catalogue of known software for Linux), and there does not
appear to be a binary package of PCCTS for Linux available, although I found 
that most versions, including v1.33 have compiled out of the box for me 
(albeit with occasional warnings).  I would also like to put the product in 
the contrib directory at the PCCTS site (if you are agreeable).  

Anyway, my questions are:

Which site would you prefer me to list as the primary site (i.e. sunsite or
ftp.parr-research.com)?  
Would you like me to undertake a binary and source distribution of PCCTS for 
Linux to upload to sunsite?  
And last but not least, if time permits (as I don't expect it will) do you 
have any suggestions for improvement of the grammar or the supporting code,
either for the current version (which is by way of being a prototype), or 
for the future version which I propose to start work on when the first phase 
of my project is completed?

Regards and thanks again for PCCTS


Tim Littlefair