Related articles |
---|
compiler/translator testing techniques ruslanv@cadence.com (Ruslan Vorobey) (2003-03-30) |
Re: compiler/translator testing techniques cdg@nullstone.com (Christopher Glaeser) (2003-03-30) |
Re: compiler/translator testing techniques ruslanv@cadence.com (Ruslan Vorobey) (2003-03-31) |
Re: compiler/translator testing techniques scott.moore6@attbi.com (Scott Moore) (2003-04-05) |
Re: compiler/translator testing techniques cdg@nullstone.com (Christopher Glaeser) (2003-04-05) |
From: | "Christopher Glaeser" <cdg@nullstone.com> |
Newsgroups: | comp.compilers |
Date: | 5 Apr 2003 15:13:57 -0500 |
Organization: | Concentric Internet Services |
References: | 03-03-160 03-03-193 03-03-202 |
Keywords: | testing, |
Posted-Date: | 05 Apr 2003 15:13:57 EST |
> The question concerned ... approaches to verify ready compiler
> created with lex/yacc pair or fully custom how it conform to
> customer's requirements.
Is this a completely new language or an extension of an existing
langauge?
Some years ago I developed the MPL compiler for Maspar, a SIMD
language based on C and the GNU technology. To test the compiler, I
used an existing commercial test suite and then wrote many thousands
of test cases using the test suite as a framework and template. This
approach proved very successfull and the final compiler product was
quite stable, particularly the SIMD constructs, but it took many
months to write all the test cases. The test development was the
dominate effort in the project, with very little time devoted to
actual compiler development by comparison. The project would not have
been possible without the original C test suite as Maspar lacked the
resouces to develop a test suite from scratch.
If you are defining a completely new langauge, I believe Infinite
Monkeys offers an automated test generation system that might prove
useful.
Hope that helps.
Best,
Christopher
Return to the
comp.compilers page.
Search the
comp.compilers archives again.