Login  Register

Re: reliability of macros

Posted by Martin du Saire on Nov 28, 2006; 7:54pm
URL: http://imagej.273.s1.nabble.com/reliability-of-macros-tp3700932p3700934.html

If I understand the question, the reliability of a macro is
determined (a) by how accurately it does arithmetic, (b) the
properties of the analytic method being implemented (Sobel edge
detect, for example), (c) the actual implementation of the method.  I
would guess that the arithmetic of the macro is as good as Java can
provide (correct?).  See

http://www.amazon.com/Java-Number-Cruncher-Programmers-Numerical/dp/0130460419

for a discussion on numerical methods in Java.

Once implemented the macro should give the same result with the same
image, "every time".  Is that what you mean by reliability?  How well
it works with a group of "similar" images is a whole separate problem
that you will have to figure out by testing the macro.

M

At 11:31 AM 11/28/2006, you wrote:

>Hi Paul,
>
>I'm not sure what you are looking for here. Statistics like "macros
>are 90% reliable" are vague and essentially meaningless. If you want
>assurances that the macro language does not contain bugs, of course
>that cannot be guaranteed, but know that many people use macros every
>day with very good results. They typically execute as expected, and in
>general are very robust.
>
>And if there are specific bugs, the ImageJ community is both extremely
>active, and led by a very talented and responsive developer who
>releases new versions on a rapid development cycle, so you can feel
>confident that bugs in the macro language will be fixed very quickly.
>
>I think the biggest gotcha for new ImageJ macro programmers is that
>the macro recorder does not always record every action taken (for
>example, recording brightness/contrast adjustments does not work as
>expected). But there is quite a bit of documentation about the macro
>language on the web site to assist with puzzling things out, and if
>that's not enough there's always the mailing list.
>
>What more do you want?
>
>-Curtis
>
>On 11/28/06, Paul Batty <[hidden email]> wrote:
>>G'day folks,
>>
>>I was just wondering how reliable are macros? I know this was
>>touched upon in previous listings saying that they are very
>>reliable, but no-one has explicitly given a figure - 90%? 99%??? I
>>have never seen anything in the literature stating anything, yet
>>everyone uses them! surely there must be some source of error -
>>nothing can really be perfect, or can it.......
>>
>>can anyone 'put this one to bed'
>>
>>Regards
>>
>>Paul.