-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
Brent,
[You wrote]
>From a security standpoint I agree with you, but if I
>was working on a hardened configuraion I wouldn't
>install precompiled software at all.
>
>In fact, I might not even use Poplog if it will only
>bootstrap via an existing Poplog compiler. You've no
>doubt seen (IIRC) Kevin Ritchie's hack to a C compiler
>to recognize when telnet was being built and insert a
>back-door...
>
I will have to go and have a look.
Since Poplog is largely written in Pop-11, it would be comparatively
easy to do something similar. The content-verification issue is
something we are going to have to address.
>Debian uses a format called "deb" that performs a
>similar role to RPM. The RPM and Debian people have
>been collarorating off-and-on for a while on a unified
>format, but I don't think much real progress has been
>made recently.
>
I assume that with both, you point a tool at a directory and say, build
with everything below this point? Is there any verification mechanism,
against a previous build or database, to at least log new or missing
files?
>
>Short Intro:
>
Thank you.
I'm a software engineer with my own (very) small firm. I tend to work on
n-tier systems, against a variety of relational database management
systems, with C/C++/TCL on the server and, most frequently, VB on the
client. At times, I've been working on Graham Technology's GT-X
software, scripting front-ends and writing or supporting service
objects, or improving installations' administration scripts. I get to do
a lot of shell scripting, lots of database interfaces, and endless SQL
(although that usually means a missing index).
I am told that many of my fellows like to specialise on either the
server- or the client-side of development, hence I get demand from a
wide range of industries.
I've always liked to play with user interfaces, operating systems,
databases and programming languages and nearly used Poplog for my final
year undergraduate project, many years ago, before plumping for DEC-10
Prolog. I got to play with Pop soon after graduating and, amongst other
things, developed a little graphical version of a network editor, using
Pop-11 with the X toolkit, based on original work by John Kellett who
used SunCORE and SunVIEW. Other than this, my experience as a Poplog
user is quite limited.
I have similar concerns about the quality of software and difficulty of
managing a source base. I make extensive use of generators, operating on
database schemas, record-type definitions, protocol packet schemas,
etc., and I have recently begun investigating practical, simple,
portable refactoring and restructuring tools that I can throw at any
project.
Many of my clients have a constantly changing set of requirements, and I
need to be able to manipulate the code case to meet those requirements.
"Programming" is not always the solution - manipulating a code base,
like any other data-type, possibly may be. All those implicit
relationships in the code base need to be rendered explicit, and then
they can be managed.
I'd like to see source code tagged, so that I can hide the annotations
when they get in the way, work on just the code that matches a specific
sub-set of compiler directives, and add links to requirements, while the
editor silently manages explicit links to match all the implicit ones I
create every time I add another token to the source stream. I currently
see much of the problem domain as a set of inter-related name-spaces
linked by mapping functions, and a source editor and linguistically
expressed code don't give me the control I crave. Describing a network
using language takes so damned long and is a pain. I want to construct
and manipulate it, and limit the use of language to the queries I use to
express the subset of the information I need at any given time. But
then, except when writing, I sometimes tend towards the autistic and/or
introverted, so perhaps this is just my little problem!
Poplog has potential for providing me with the cross-platform tools I
want to apply to a wide range of languages and environments. It is
particularly appropriate for linguistic manipulation and has a variety
of parsing and list manipulation functionality as standard. A universal
parser is just a general mechanism, maintaining networks of name-spaces
and a collection of specific data sets, with recognisers and
manipulators feeding off the data to support intelligent editing,
querying, syntactic and semantic checking, long before the compiler is
required to be fed a single morsel of code. Or so I hope. (To date,
almost all of my experiments have used VB which, due to
over-familiarity, I can throw onto a screen faster than any other
language I type. I'd like to change that.)
Regards,
- --
Jeff
-----BEGIN PGP SIGNATURE-----
Version: PGPsdk version 1.7.1
iQA/AwUBQD/Tk/Hj+enJbeYqEQKAegCfZ4Wd3AbPS45bx9Fj5QM8ExGj1nQAn1kh
/KGUZmX8tdO+X+lBRjkp2P6C
=Sqky
-----END PGP SIGNATURE-----
|