public inbox for gcc@gcc.gnu.org
 help / color / mirror / Atom feed
From: "Sławomir Lach" <slawek@lach.art.pl>
To: gcc@gcc.gnu.org
Subject: can-be-null can-not-be-null break-instruction for better handling pointers.
Date: Thu, 27 Apr 2023 17:54:52 +0200	[thread overview]
Message-ID: <13261922.uLZWGnKmhe@localhost.localdomain> (raw)

I am not C expert, so be polity. I do not see something similar in C world, 
but similar techniques in other languages, such like Vala.

I suggest to create two new pointer type:
1. can-be-null
2. cannot-be-null
(You must find other words to describe it behavior).

First enforces to compiler checking it is not null, when it was used (by -> 
operator) or used, when cannot-be-null is excepted. Of course, user can cast 
to cannot-be-null to avoid checking. So:
1. Usage with -> operator, or * must be placed inside conditional block, with 
contains null check in condition
2. Passing as cannot-be-null (except explicit conversion) will requires to put 
in block as above

The same restrictions apply to put normal pointer in case, when cannot-be-null 
excepted.

Additionally, break-instruction word will be reserved for functions, so exit 
will be traded as end of control-flow and further code will be traded as 
checked.

Imagine assert will be traded as break-ins

This code:

int get_vector_size(struct vector cannot-be-null *vec)
{
   assert(vec != NULL)

   return vec->size;
}

Will be correct, because assert macro will (possible) be extended to:

if (!(vec != NULL)) {
    puts("assertion error: vec != NULL");
    exit(1);
}

And exit will be break-instruction.

We could use cannot-be-null in some cases, such like:

int year;

scanf("How old are you:%d", &year);

But user could pass NULL as second parameter.

When pointers in scanf parameters are cannot-be-null, the compiler will 
disallow pass null by us.

Of course - inside scanf must check if each parameter are not NULL.

Another code:

void error(const char *message)
{
   perror(message);
}

And perror will have cannot-be-null word describe one's parameter.
This will cause compile-time-error, because message was passed to function, 
which parameter is cannot-be-null.

Programmer should do:
void error(const char *message)
{
   if (NULL != message)
     perror(message);
}

What do you think. This will solve a lot of problems,

It could be also great to introduce function checking if any of it parameter 
are null, but I do not known how to make it good-quality (simple in usage). 
Programer should propably pass number of passed parameters, or type of 
parameter before each.



             reply	other threads:[~2023-04-27 15:54 UTC|newest]

Thread overview: 2+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2023-04-27 15:54 Sławomir Lach [this message]
2023-04-28 21:34 ` Christopher Bazley

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=13261922.uLZWGnKmhe@localhost.localdomain \
    --to=slawek@lach.art.pl \
    --cc=gcc@gcc.gnu.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).