Re: TenDRA

Robert Andrews <rca@rivers.dra.hmg.gb>
18 Feb 1998 23:04:55 -0500

          From comp.compilers

Related articles
TenDRA luc.tanguay@bell.ca (TANGUAY, LUC) (1998-02-14)
Re: TenDRA fabre@gr.osf.org (Christian Fabre) (1998-02-18)
Re: TenDRA rca@rivers.dra.hmg.gb (Robert Andrews) (1998-02-18)
TenDRA foskaty@my-deja.com (Ole Foskaty) (1999-10-27)
Re: TenDRA andyj@mc.com (Andy Johnson) (1999-10-28)
Re: TenDRA Roger@natron.demon.co.uk (1999-10-29)
Re: TenDRA broonie@tardis.ed.ac.uk (Mark Brown) (1999-10-31)
Re: TenDRA bruce@cenderis.demon.co.uk (Bruce Stephens) (1999-10-31)
Re: TenDRA bruce@cenderis.demon.co.uk (Bruce Stephens) (1999-11-01)
[3 later articles]
| List of all articles for this month |
From: Robert Andrews <rca@rivers.dra.hmg.gb>
Newsgroups: comp.compilers
Date: 18 Feb 1998 23:04:55 -0500
Organization: DERA
References: 98-02-065
Keywords: tools, UNCOL

LUC TANGUAY wrote:
> Hi, is there anybody out there having some experience with TenDRA ?
> Could it be used to analyze and/or visualize C/C++ source code ? And
> difficult it is to build a producer for a Pascal-like language ?


There are a couple of commercial C source code visualisers based on
the TenDRA C front-end available, namely CodeIntegrity by MKS and
OpenLint by Softway. Both of these use a simple symbol table dump
generated by the C producer rather than the ANDF intermediate form.
The C++ producer can also dump its symbol table in the same way, so
generating similar tools for C++ would also be possible.


There has been some work here at DERA (not the TenDRA people, but
another group) on doing flow analysis more directly on the generated
ANDF. (Luc, I'll email you a contact name when I've discussed it with
the people involved.) The structured nature of the ANDF makes it a
convenient form for analysis; the chief problem is in relating the
gleaned information back to the original source code, because most of
this information (line numbers, variable names etc.) has been thrown
away.


The information can however be preserved using the -g option to
include diagnostic information in the ANDF. Note that the current
(ANDF 4.0) diagnostic specification only really covers C (diagnostic
information is far more language-specific than code); we are currently
working with DDC-I to develop a new ANDF-DE (ANDF diagnostic
extension) which also covers C++ and Ada. There is experimental
support for an early version of this format in the current release
(use the -g2 or -g3 option to tcc).


I don't know of anyone who has looked at doing a Pascal producer. My
experience doing the C++ was that 95% of my problems were C++ problems
(i.e. problems I would have had with any C++ compiler), and only 5%
were ANDF problems.


- Rob
-------------------------------------------------------------------
Rob Andrews
DERA (Defence Evaluation and Research Agency)
Email: R.Andrews@eris.dera.gov.uk or ra@rivers.dra.hmg.gb
http://alph.dera.gov.uk/TenDRA/
--


Post a followup to this message

Return to the comp.compilers page.
Search the comp.compilers archives again.