From: Junio C Hamano <gitster@pobox.com>
Date: Mon, 08 Apr 2024 20:36:40 GMT
Subject: Re: [PATCH] reftable: use xmalloc() and xrealloc()
Message-ID: <xmqqzfu3a8x3.fsf@gitster.g>
In-Reply-To: <CAOw_e7Z9dGeVU399D6o37L3am0abnYUrZnNQEFKhyUv=A2=j8g@mail.gmail.com>

Han-Wen Nienhuys <hanwenn@gmail.com> writes:

> However, it is probably pointless as long as strbuf_* functions do not
> signal OOM gracefully. There was some talk of libifying strbuf. Did
> that work include returning OOM error codes in case malloc returns
> null? A quick look at strbuf.h suggests not.

I would expect not.

The "libified" strbuf (aka "strbuf API in the Git std lib") will
have to be different from what we internally use from <strbuf.h>.
<gitstdlib/strbuf.h> will export gitstdlib_strbuf_addstr(), which is
"properly" libified and signals an allocation failure to its caller.

When that happens, I would expect that strbuf_addstr() would be a
thin wrapper around gitstdlib_strbuf_addstr(), and still just dies
with "we ran out of memory", i.e.

	/* strbuf.h */
	#include <strbuf.h>
	#include <gitstdlib/strbuf.h>

	void strbuf_addstr(struct strbuf *sb, const char *s)
	{
		int err = gitstdlib_strbuf_addstr(sb, s);
                if (!err)
			return; /* happy */
		switch (err) {
		case GITLIB_OOM: /* there may be others */
			die("Out of memory");
			...
		}
	}

which would keep the damage to Git codebase to the minimum when we
become the first client of the "Git std lib".