SVR4 Languages SIG

Dan Oldman <oldman@DG-RTP.DG.COM>
Wed, 5 Sep 90 9:11:02 EDT

          From comp.compilers

Related articles
SVR4 Languages SIG oldman@DG-RTP.DG.COM (Dan Oldman) (1990-09-15)
| List of all articles for this month |

Newsgroups: comp.compilers
From: Dan Oldman <oldman@DG-RTP.DG.COM>
Organization: Compilers Central
Date: Wed, 5 Sep 90 9:11:02 EDT

Aug. 27, 1990
Call for Participation:


UNIX System V Programming Language issues SIG


Announcing the formation of a UNIX International Special Interest Group on
Programming Language Issues. This group will act as a clearing house for UI
member companies and other interested parties to resolve issues of supporting
various programming languages on UNIX System V. One pressing problem is the
support of debugging.


Debugging Support
=================


One significant change introduced by System V Release 4 is the replacement of
COFF (Common Object File Format) with ELF (Executable and Linker Format)
representation of programs. COFF, despite many problems, had a barely
acceptable but functional representation for debugging information. ELF, at
this time, lacks anything but the suggestion that the .debug section might
contain some debugging information. There is also only a weak standard in the
area of debugger interface to the kernel.


The lack of a standard is a serious impediment to third party compiler
writers who wish to work with the standard system debugger and with third
party debugger writers that wish to operate on many UNIX platforms with
standard and third party compilers. Any programmer who wishes to debug an
application that is built with two or more different compilers is also hurt
by the lack of a standard.


Attempting to develop standards in this area is not new. The needs of the
popular programming languages and debuggers are substantial and past attempts
have ended with the current situation of "agreeing to disagree". This ignores
the fact that the 80-20 rule applies well here. We can solve the needs of 80%
of the community with 20% of the work. If we design extensibility into the
standard, the remaining 20% of the community can build on top of the standard
and get more done with less effort.


UNIX Software Labs has developed a new debugger representation, called DWARF,
that is used in SVR4 C Issue 5 compiler and SDB debugger. It has some of the
qualities of an acceptable standard and would probably be a good place to
start.


Goals of the SIG
================


As stated earlier, the overall goal of the SIG is to provide a clearing house
for UI member companies and anyone else who has an interest to resolve
programming language support related issues on UNIX System V. There are some
specific projects that the SIG must complete as soon as possible:


        1. Develop a robust and efficient framework for debugging
              information. This framework will consist of a generic base
              that deals with the common problems of the popular third
              generation languages and supports extensibility for other
              uses.


        2. Define how the following languages map onto this
              framework: C (K&R and ANSI), Fortran (77 and 90), C++, and
              probably others.


        3. Define the format of "core" files.


        4. Define the interface between the debugger and the RTLD
              (shared library runtime loader).


        5. Define a standard for Kernel support of debugging and then
              extend that standard to deal with upcoming features like
              threads.


Beyond that there are other issues that could be standardized such as support
for debugging in the absence of debugger information and support for long long
integer types. I'm sure that there are more things that I have not mentioned
here.


Mechanism
=========


It is the intention of the SIG to have bi-monthly meetings and heavy network
and conference call communication. We plan to have an organizational meeting
in late September. Here's a start for the proposed agenda. I will add
anything else to it that is requested. (See feedback below.)


        Introductions and opportunity for individuals to state their
interest and what they expect the SIG to accomplish


        Charter discussions and election of officers


        Standardization priorities


        Debugger issues


              Overall Goals of Debugging Information


              Proposal for the framework


        Discussion about the next meeting


Feedback
========


For the first meeting, I am targeting Thursday September 27th in the Boston
area, but a final decision won't be made until I hear from the people who are
interested in being involved. Please respond before Friday, September 7th,
via email, fax, mail, or phone with the information requested below. I will
acknowledge any messages that you send me to avoid things being lost in the
mail.


        Name:
        Organization:
        Area of interest:
        Phone number:
        Fax number:
        Email address:
        Mail address:
        Level of interest:
              __ I will attend most meetings and am willing to do
                    significant spec writing.
              __ I will attend most meetings and actively review
                    proposals.
              __ I will actively monitor the email discussions, but will
                    not be able to attend meetings.
              __ I am interested in occasional status reports.
        Preferred location for meetings:


        __ I can make it to the first meeting if it adheres to the
              following time or location constraints:


        I would like the following added to the agenda:


        Any other comments:




Thank you for your interest.
------------------------------------------------------------------
Dan Oldman internet: oldman@dg-rtp.dg.com
Data General Corporation uucp: ...!mcnc!rti!dg-rtp!oldman
62 Alexander Drive voice: (919) 248-6125
Research Triangle Park, NC 27709 fax: (919) 541-9089
--


Post a followup to this message

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