Re: Compiler stress tests?

jeffncyn@internetmci.com (Jeff & Cynthia Enderwick)
12 Jan 1997 11:30:50 -0500

          From comp.compilers
Related articles
Compiler stress tests? pardo@cs.washington.edu (1997-01-02)
Re: Compiler stress tests? d.sand@ix.netcom.com (Duane Sand) (1997-01-03)
Re: Compiler stress tests? cliffc@risc.sps.mot.com (1997-01-03)
Re: Compiler stress tests? clyde@hitech.com.au (1997-01-04)
Re: Compiler stress tests? jeffncyn@internetmci.com (1997-01-12)
Re: Compiler stress tests? gah@u.washington.edu (1997-01-16)
Re: Compiler stress tests? stephen.baynes@soton.sc.philips.com (Stephen Baynes) (1997-01-17)
Re: Compiler stress tests? jch@hazel.pwd.hp.com (John Haxby) (1997-01-22)
| List of all articles for this month |
From: jeffncyn@internetmci.com (Jeff & Cynthia Enderwick)
Newsgroups: comp.compilers
Date: 12 Jan 1997 11:30:50 -0500
Organization: InternetMCI
References: 97-01-020 97-01-024 97-01-033
Keywords: C, testing

Clyde Smith-Stubbs wrote:
>
> It's quite frightening when you apply this kind of verification
> to an existing compiler - we find it's normal that the test suite
> has only been testing about 70% of the compiler! Of the balance,
> some turns out to be unreachable, much proves to be correct, but
> some is bound to be wrong.


When I used to write compilers we used Plum-Hall, Perennial, and the
usual kitchen sink. Using gct, I discovered that many modules of the
compiler would only get ~50% *execution* coverage, never mind branch
coverage.


We did pretty well w.r.t. bugs by being careful, unit testing during
development, and adding source that broke the compiler to the test
suite.


Jeff
--

Post a followup to this message

Return to the comp.compilers page.
Search the comp.compilers archives again.