From octave-maintainers-request at bevo dot che dot wisc dot edu Sat Apr 21 11:27:32 2001 Subject: Re: request for "assert" From: "Joseph P. Skudlarek" To: pkienzle at kienzle dot powernet dot co dot uk CC: octave-maintainers at bevo dot che dot wisc dot edu Date: Sat, 21 Apr 2001 07:58:16 -0700 Thanks for the reply, Paul. Ah, I didn't explain myself very well. Actually, I was thinking of something even easier, an assert routine which takes a boolean value (most likely generated by a conditional expression), and if the value is not true, it simply errors out. The version I'm using goes like this: ## -*- Mode: octave; indent-tabs-mode: nil fill-column: 72 -*- ## function assert(assertion) if (! assertion) error("assertion failed"); endif endfunction ##[] I use it in cases like this: ... bin_width_hz = 1/Ts / w; bin_3dB = 500e3 / bin_width_hz; assert(bin_3dB < r); ... What you described I've seen MFC refer to as "AssertValid", rather than just assert. What I'm think of is more like "assert" from the C library. If assert were an Octave builtin, then it potentially could also state the test of the conditional which failed, like I get from my C compiler: a.out: assert.c:5: main: Assertion `0 == 1' failed. I hope this helps. Thanks for helping to maintain a wonderful! tool. /Jskud >------ Begin Included Message ------ > Date: Sat, 21 Apr 2001 13:24:32 +0100 > To: "Joseph P. Skudlarek" > Cc: octave-maintainers at bevo dot che dot wisc dot edu > Subject: Re: request for "assert" > Content-Disposition: inline > User-Agent: Mutt/1.2.5i > From: Paul Kienzle > > I've got a version at: > > http://users.powernet.co.uk/kienzle/octave/matcompat/scripts/testfun/assert.m > > It recursively walks structure trees, but doesn't handle lists. Feel free > to extend it. > > Ideally, each type should provide a comparator which determines if two > instances of that type have the same value within some error bound so > that assert would also work for user defined types. Add an isequal > method to the octave_value class? > > Paul Kienzle > pkienzle at kienzle dot powernet dot co dot uk > > On Fri, Apr 20, 2001 at 09:22:31PM -0700, Joseph P. Skudlarek wrote: > > Hi. An assert is a wonderful thing. It's easy for a user to add, but > > given the utility and the benefit, it might be better if it were built > > in -- more people might use it. So, how 'bout considering it? Thanks. > > /Jskud > > > > > >------ End Included Message ------