comp.lang.ada
 help / color / mirror / Atom feed
From: Jacob Sparre Andersen <jacob@jacob-sparre.dk>
Subject: Re: Ada in Denmark: Birthday of Ada - Talk on cyclomatic complexity
Date: Wed, 11 Dec 2013 12:32:47 +0100
Date: 2013-12-11T12:32:47+01:00	[thread overview]
Message-ID: <87zjo7r5b4.fsf@adaheads.sparre-andersen.dk> (raw)
In-Reply-To: 87r49lhwh2.fsf@adaheads.sparre-andersen.dk

A short report from the Ada in Denmark meeting yesterday:

We started the evening with Thomas Pedersen's presentation of the
principles in McCabe's cyclomatic complexity measure.  Following up on
the presentation we discussed the experimental evidence available to
identify which (cyclomatic) complexities are acceptable.  Per Dalgas was
kind enough to provide a "problematic" legacy subprogram in Ada which we
found to have a cyclomatic complexity of 9.  McCabe himself indicates 10
as an upper limit and references some provided examples of problematic
subprograms with complexities of 16 and above (IIRC).  I have done a
search for Ada subprograms with complexity above 10 in my published Ada
projects.  All the examples I have found so far were straight
transcriptions from old FORTRAN sources.

The discussion continued on the subject of measuring source text
quality.  Per Dalgas posed the challenge of how we get more software
developers to _use_ the available metrics.  One option which came up was
to run quality metrics automatically on the source texts on public
version control repository services (such as Bitbucket, Github and
Sourceforge).  This has the benefit of being something which initially
can be implemented as an independent service and only later be pushed to
the actual source hosting services (if they want it).  But would
implementing such a measure make a difference?  Will the developers
worry about it?  Will the users of software use it as a selection
criteria if they get the possibility?

Then SQALE came up as an example of a "combined" source text quality
measure (I had the pleasure of attending Jean-Pierre Rosen's talk on
SQALE at Ada Europe 2011), but it appears that only few tools exist (and
none of them Open Source) and it wasn't obvious which languages the
tools can analyse.

In another branch of the discussion we wondered if we could get big
buyers/tenderers of software to require SQALE measures with constraints
on what is acceptable levels as a part of software delivery contracts.

Greetings,

Jacob
-- 
"Very small. Go to sleep" - monster (not drooling)


  reply	other threads:[~2013-12-11 11:32 UTC|newest]

Thread overview: 3+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2013-12-10  9:43 Ada in Denmark: Birthday of Ada - Talk on cyclomatic complexity Jacob Sparre Andersen
2013-12-11 11:32 ` Jacob Sparre Andersen [this message]
2013-12-12  7:42   ` Ludovic Brenta
replies disabled

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox