Re: question about linker

On 04/12/2024 19:23, Janis Papanagnou wrote:

On 04.12.2024 18:43, Bart wrote:

[...]

>
Do you use a formal grammar when parsing a CSV file, or something
equally trivial?

 Is that the reason why there's so many versions around that are
incompatible? CSV-parsing is not "trivial" if you look into the
details; you have to specify these (at first glance not obvious
details) to be sure that your "CSV-data" works not only with your
"CSV-parser" but that there's a common understanding of the CSV-
"language". (It's only few details; delimiters in string values,
escapes, and such, but enough to initiate incompatible formats.)
Yes, of course; if there would have been a formal specification
in the first place we wouldn't have the mess we now actually have.
 And if you anyway write your tools only for yourself, and if you
don't intend to exchange data with others, no one cares what you
think a/the "CSV-format" actually is.
 But we weren't discussing such comparably simple structures; we
have been discussing programming languages (and their grammars).
And most of us are considering sensible languages, not privately
hacked up toy languages or implementations of personal hobbies.

You really hate toy languages don't you?
The fact is that a compiler is only a bit of software like anything else. It might take some input, process it, and produce output.
When someone has claimed to write some program, do you always demand they produce a 'formal grammar' for it?
What is the complexity threshold anyway for something to need such a formal specification?
What input would even be classed as a language: are binary file formats languages? What about the commands of a CLI?
More importantly, why are some people here so supercilious, condescending and patronising?
I don't create formal grammars for everything for the same reasons I don't file a flight plan every time I drive to the supermarket.
I'm starting to suspect that the few details that have emerged of my languages' module scheme have shown it is actually quite decent. So let's find something else to belittle Bart with!

Human languages (specifically English with its many irregularities)
are worse than a formal language and a unreliable base and unsuited
as programming languages.

PE file is described in English. Most file binary file formats are.

That's nothing more then a hacker's feeble excuse to justify his
ignorance.

So you're calling me a hacker, and you're calling me ignorant.
Nice.

>
Or the parser was created for a trivial language and has evolved.

 Like, as I've heard, the Unix shell, with all it's irregularities
and quirks? - You really think this is a good paragon?
 Even if some hacker defines a language ad hoc - and Intercal comes
to my mind - for a serious programming language you should have
[documented] syntax and semantics, and why the hell would anyone
use English instead of a formal specification for syntax.

When YOU devise a language, then YOU can choose to do it whichever way YOU like. I wouldn't care one way or another.
My own products were in-house tools to get stuff done. And they worked spectacularly well.

A formal or informal grammar might be useful in a language reference -
once the language is stable.

 For a hacker like you maybe. Professionals typically specify before
implementation.

Again. I bet the first C compiler didn't bother with a grammar!
What is it with people here?
OK, you're a professional; I'm an amateur (an amateur who managed to retire at 42). I will give you that. Satisfied?

 

[...]
>
For the rest of us, that part is the simplest part of a compiler. You
write it, and move on.

 Is that the reason why your languages and compilers are so widespread
used? </sarcasm>

By contrast, the languages that you've devised are in common use of course?