|Incorporating run-time error handline in code generation jdm1intx@DIE_SPAMBOT_DIEairmail.net (jdm) (2001-12-20)|
|Re: Incorporating run-time error handline in code generation email@example.com (Paolo Bonzini) (2001-12-22)|
|Re: Incorporating run-time error handline in code generation Arargh@Arargh.com (Arargh!) (2001-12-22)|
|Re: Incorporating run-time error handline in code generation firstname.lastname@example.org (2001-12-22)|
|From:||"Paolo Bonzini" <email@example.com>|
|Date:||22 Dec 2001 22:57:29 -0500|
|Organization:||Mailgate.ORG Server - http://www.Mailgate.ORG|
|Posted-Date:||22 Dec 2001 22:57:29 EST|
> For example, the various flavors of Microsoft Basics have the "On
> Error Goto" construct and the Err object. How much of that is part of
> the runtime library, and how much is part of code generation ? Is
> there a website or book that has more comprehensive information on
> this subject?
Languages with access to execution contexts as objects
such as Smalltalk can do it entirely with code in that language
-- but probably to this level we're cheating because in practice
it is the runtime is making things available at a very high level.
The Java VM reference probably provides a more low level
treatment of the subject.
If that's not enough, source code for gcc will let you know what the
runtime library expects from the code generator (gcc's runtime library
is sensibly less hairy than the rest of the code). In turn gcc is
based on the public IA64 ABI for C++ (which the gcc folks extended to
Win32 has a cross-language API for exception handling which compilers
are supposed to feed with information generated at compile time; the
reference for this API will let you know this.
DDJ's November 2001 issue has an article on how Kylix (Delphi for
Linux) does its exception handling.
Return to the
Search the comp.compilers archives again.