[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [Tinycc-devel] Question about in-memory compilation target
From: |
Steffen Nurpmeso |
Subject: |
Re: [Tinycc-devel] Question about in-memory compilation target |
Date: |
Fri, 09 Feb 2024 00:24:58 +0100 |
User-agent: |
s-nail v14.9.24-596-g7894190075 |
Sean Conner wrote in
<20240208102201.GB22657@brevard.conman.org>:
|It was thus said that the Great grischka via Tinycc-devel once stated:
|> On 07.02.2024 09:38, Eric Raible wrote:
...
| Lua contains a way to set a custom realloc() function (via
|lua_newstate()), thus allowing someone who is embedding Lua into an
|application to use a custom allocator, or even just restrict the amount of
|memory Lua can use. There is also a function in Lua to obtain the \
|allocator
|function Lua is currently using (via lua_getallocf()).
I think the possibility to replace an allocator is a good thing.
I was frustrated that one of the first things of the libressl fork
was to remove that possibility (i had found lots of problems in
the MUA i took maintainership of by hooking that thing), and such.
I loved the old libraries, graphics and such, which per-se
supported hooking their allocators.
Then again, today, so many pieces of cake install on-fork, etc,
handlers, use so-called sophisticated linker techniques, that
whatever you do you have a bunch of things using either the
standard allocator, or their very own, non-hookable thing, that
you are total at odds, anyhow.
As a nice counter-example, just this/last week FreeBSD introduced
the split of POSIX+ systemcalls into a libsys library, and out of
the standard C library. But this is a tiny drop, and non-portable
it is of course, too.
--steffen
|
|Der Kragenbaer, The moon bear,
|der holt sich munter he cheerfully and one by one
|einen nach dem anderen runter wa.ks himself off
|(By Robert Gernhardt)