XPost: alt.folklore.computers
From: antispam@fricas.org
In alt.folklore.computers The Natural Philosopher wrote:
> On 01/01/2026 14:28, Peter Flass wrote:
>> On 1/1/26 05:49, The Natural Philosopher wrote:
>>> On 01/01/2026 03:07, c186282 wrote:
>>>> On 12/31/25 17:35, The Natural Philosopher wrote:
>>>>> On 31/12/2025 19:21, c186282 wrote:
>>>>>> I've writ stuff with five or six levels of nesting
>>>>>> but don't like it, usually if/then/else stuff. Oft
>>>>>> re-did it later to be more easy to follow. IMHO
>>>>>> readability/comprehensibility is as important as
>>>>>> functionally correct code.
>>>>>
>>>>> 100% agree.
>>>>>
>>>>> Often write little functions that are only called once. Merely to
>>>>> lexically separate atomic functional blocks.
>>>>>
>>>>> No idea whether the compiler/linker inlines them or not.
>>>>>
>>>>> There is nothing worse than making top level decisions followed by
>>>>> some nitty detail to detect some low level error.
>>>>>
>>>>> e.g. assume a call to allocate memory always works or the call will
>>>>> do the appropriate jump to a global error handler to abort things
>>>>> cleanly.
>>>>>
>>>>> The point of structure was supposed to be to elucidate program flow,
>>>>> not obscure it with elegant formally correct cruft.
>>>>
>>>>
>>>> Agree.
>>>>
>>>> As I've said before, I'm still quite fond of Pascal and
>>>> write apps of various size in it (oft first proto-ed
>>>> in Python). The structure is 'elegant', but you CAN
>>>> carry it TOO far, to where it gets in the way instead
>>>> of helping things.
>>>>
>>> My one and only experience of trying to make Pascal do what was
>>> trivial in 'C' led me to resolve never ever to touch it again.
>>>
>>> If you are trying to write - as it turned out I was - a disk driver in
>>> pascal, where a given sector may be a byte stream, a series of 16 bit
>>> integers, or a structure defined by thee first few bytes in the
>>> sector, you end up with a massive union that is so cumbersome it is
>>> almost impossible to read - let alone use.
>>
>> Doesn't Pascal have variant records?
>>
> IIRC it (Turbo Pascal. The amateurs language) had unions of some sort,
> but I would have needed about 100 to cover all cases and it was even
> then messy.
Turbo Pascal could do essentially all thar C could do (and do things
which were not strightforward in C, but this is irrelevant here). And
do this in a very similar way, once you knew how Turbo Pascal constructs
worked. If you really needed 100 variant record in Turbo Pascal,
then you needed 100 unions in C. If you could do this more
simply in C, you could do this more simply in Turbo Pascal too.
Given what you wrote, it looks that you simply lacked experience
writing Turbo Pascal. In other words, you were unqualified to
do the job that you were supposed to do (write the driver in
Turbo Pascal), so you decided to do thing that you know how to
do, that is to write it in C.
IMO biggest drawback of Turbo Pascal was poor speed of generated
code (and size too). For me deal breaker was fact that Turbo
Pascal was 16-bit and tied to DOS. DJGCC gave me 32-bit
integers and slightly later I switched to Linux, so Turbo
Pascal was not longer relevant for me. But if you were
programming 16-bit DOS and did not mind poor speed of generated
code, than IMO Turbo Pascal was quite decent programming
language, quite competitive in expressivity to C.
> The existing code was shit anyway so I rewrote the lot
That could be true. But given that you apparently do not know
what Turbo Pascal can do, it is not clear if you are qualified
to judge the code.
BTW: It is normal and common for programmers to want to
rewrite/write from scratch instead of understanding and
improving existing code. But in most cases working on
existing code leads to better effect.
--
Waldek Hebisch
--- SoupGate-Win32 v1.05
* Origin: you cannot sedate... all the things you hate (1:229/2)
|
