From: 643-408-1753@kylheku.com
On 2025-10-26, Michael S wrote:
>> I can't imagine why anyone would write cdecl (if it is written in C)
>> such that it's anything but a maximally conforming ISO C program,
>> which can be built like this:
>>
>> make cdecl
>>
>> without any Makefile present, in a directory in which there is just
>> one file: cdecl.c.
>>
>
> You are exaggerating.
> There is nothing wrong with multiple files and small nice manually
Yes, I'm exaggerating; of course I can imagine using more than
one file for cdecl.
I would say that if you need two files to write cdecl, and
one of them is not an accurate grammar file for a parser generator
(needing to be a spearate file due to being in that notation),
which handles things int (*p)(int (*q)(void * const x)),
you've massively fucked it up.
> In that regard autotools resemble Postel's principle - the most harmful
Postel's principle is awful, requiring paragraphs of apologetic
defense to explain what Postel really meant and how it made sense in his
context, so that it wasn't actually idiotic.
Programs should be conservative in what they generate, and loudly reject
any input that is out of spec.
Programs that accept crap are good for business, because naive
customers just see that those programs "work" with some input
that other programs "don't handle".
They are harmful to the ecosystem, creating a race for the bottom
competition in which specs fall by the wayside while programs struggle
to handle buggy inputs, and nobody knows what is correct any more.
--
TXR Programming Language: http://nongnu.org/txr
Cygnal: Cygwin Native Application Library: http://kylheku.com/cygnal
Mastodon: @Kazinator@mstdn.ca
--- SoupGate-Win32 v1.05
* Origin: you cannot sedate... all the things you hate (1:229/2)
|
