Re: Will the availability of browsers affect source code?

>With the availability of multiple cross plattform world wide web
>browsers will we see any impact on how programs are coded and
>commented?
>[...]


I do not think language design needs to change much in order to
satisfy the requirements you mention. What you are describing are
better ways of documenting a software system. I think we could use a
language for documenting software systems that would take care of the
things you mention. SGML or a SGML-based language would be great for
this kind of documentation. I recently attended a course in SGML held
by Erik Naggum (you will know him if you have been comp.text.sgml ;)
). I learned that SGML is a language designed to represent
information. SGML is the skeleton and HTML and other languages are
designed around it (one such language is HyTime and seems to be quite
interresting).


SGML and HTML have primarily been used as tools for structuring
textual data, but I think that they could equally well be used to
represent externally information that compilers used to store only for
internal use. If we could get that information out of the hands of the
compiler and into a portable format I think there are
oppurtunities. There are already several systems and compilers that
export this kind of information to browsers or for typesetting
purposes (TeX, Info, man-pages, HTML etc.). If they exported to some
sort of ``common'' DTD it would be a big win. (a DTD is the grammar
for how a particular SGML document should be written. HTML 3.0 is a
DTD. I am working on a compiler that will produce SGML, but there is
no ``standard'' DTD to follow).


The advantage of SGML is that is not designed to do anything useful on
its own. It is instead designed so that you can process the
information or convert it in different ways. Using different
conversion tools you can produce HTML documents, TeX code, CD-ROM
versions, books etc.


When we have the source code of a system (tagged with all sorts of
things) converted to SGML, we could merge this with information we get
from other sources such as pictures, animations and man-pages - but
also information like news-discussions, faqs, URL-references, version
control information etc. What we could end up with is basically to
have information on a system more systematically collected than ever
before.


>If you think I am total off base, and this was the wrong news group
>(it isn't really on target, but in my warped twisted mind it seemd
>kinda close) then feel free to answer via email.


While I think much of the hypertekst linking and production of what
you describe lie in the domain of other tools, I think compilers
should be more cooperative and share their information ;)


astor
--

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