On 2022-09-06 13:54, Arthur Schwarz wrote:
After a while I decided, why not. I might just as well answer you. Thanks for
your response. This will be a summary and not a point-to-point answer. It is
clear that Bison does not want to, and will not, proceed forwards, and so,
take this as me just grumbling in my cave.
I do think the effort is moderately difficult. I do not believe it is as hard
as you have indicated. The languages you generate Bison for is modest, and the
semantics are similar (a caveat here since I don't know D). If you don't want
to develop you're own parser, then you can always fiddle with the gcc
(https://gcc.gnu.org/) front end to get a desired result.
The problem is that there is no C, D, Java or anything complete to parse until
*after* Bison has done its job. Only then do you have a complete file that you
can feed to some front-end to analyze for dependencies or what have you.
I thought that the whole point of this proposed exercise is to determine how to
put the program together from fragments in the first place.
The way Bison works is that it needs to generate the correct, buildable output
in a single pass.
Speaking of passes, one thing you could theoretically do is treat this as a
search problem. There are a finite number of pieces going into the grammar
file, and they can be permuted in a finite number of ways, which could be
searched for a permutation that results in something that compiles.
There is the problem that during development, the code may be wrong, so that no
permutation of that code combined into a parser source file will compile. Our
permutation search loop is then sent on a time-consuming fool's errand that
comes up with nothing.
This is an important point: Bison can generate an output file even when the C,
D, Java, ... snippets have errors in them. Those are then compile- or run-time
errors for the language downstream.
Under this search idea, integration with the build would be a sticking point,
because how the code is compiled varies with the application's build
environment. Bison output for ostensibly the same language, like C, can be
built with different compilers in different environments. That could be
addressed with some sort of hooks for build integration, which will likely be
clunky to use.
When you think about it, a tool like Yacc should have a simple specification:
most of the code snippets in the second language should appear in the output in
the same order as they appear in the grammar file. Except for situations when
the order is unspecified (in which case the programmer is advised not to depend
on it). On top of that there can be some mechanism to influence the order, like
the named sections. That's it.
"... fragile magic ..." Ahem.
I suspect Akim may have been referring to what had been actually tried, in reference to
named code blocks being developed to "depart from a model where Bison 'guessed'
where to issue the user's code based on some heuristics".