[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Emacs and glibc malloc API change

From: Paul Eggert
Subject: Re: Emacs and glibc malloc API change
Date: Thu, 21 Jan 2016 21:18:44 -0800
User-agent: Mozilla/5.0 (X11; Linux x86_64; rv:38.0) Gecko/20100101 Thunderbird/38.5.1

Richard Stallman wrote:
   > So far as we know, recent Emacs should work even after the proposed glibc 
   > is installed. For example, if you build Emacs 24.5 on a system with the new
   > glibc, it should still work.

How is that?

Emac's 'configure' script probes the system for its malloc API. If 'configure' sees the special glibc malloc API, it normally uses that API; otherwise it falls back on a couple of alternatives that do not involve a special malloc API. One alternative is to use the standard malloc API without GNU extensions; another is to use a malloc implementation that Emacs provides, a modified copy of older glibc code that mostly supports the glibc malloc API. Currently the second alternative is what Emacs 'configure' will choose on a hypothetical new glibc that would lack the special malloc API.

Even without any glibc change, the current approach was reaching its limits. For example, even though Cygwin is based on glibc, its malloc doesn't work with Emacs and so Emacs 'configure' has special-case code for Cygwin that jumps through hoops that almost no one understands. Removing all this gorp from Emacs will be a win in the long run, though it'll take some work to code and test. But in the meantime Emacs should still work without change.

reply via email to

[Prev in Thread] Current Thread [Next in Thread]