[Openvas-devel] [Openvas-commits] r5228 - in trunk/openvas-manager: . src src/tests

[Matthew Mundell]

> On Friday 25 September 2009 11:13:56 Matthew Mundell wrote:
> > > ===================================================================
> > > --- trunk/openvas-manager/src/manage.h	2009-09-25 07:32:08 UTC (rev 5227)
> > > +++ trunk/openvas-manager/src/manage.h	2009-09-25 07:45:23 UTC (rev 5228)
> > > @@ -30,8 +30,8 @@
> > >  #include <glib.h>
> > >  #include <ossp/uuid.h>
> > >
> > > -#include <openvas/certificate.h>
> > > -#include <openvas/nvti.h>
> > > +#include <openvas/base/certificate.h> /* for certificate_t */
> > > +#include <openvas/base/nvti.h> /* for nvti_t */
> >
> > Please can we refrain from doing this.  It adds maintenance with little
> > gain.  It makes the include block text-heavy and messy, and it's hard to do
> > consistently.  There's little need for it in these cases.  For nvti, for
> > example, the word nvti only occurs twice in manage.h, so it's easy to see
> > the reason that the include is present.  nvti.h exports a small interface
> > anyway, so it would be pretty obvious even in a large file like manage.c.
>
> we have numerous includes all over the code without a need.
> People tend to not care about removing includes when removing code.

Are extra includes such a problem?

> The way I marked the include, my editor shows me quickly
> whether this function is still in the code or not.
> If not, I can remove the include, recompile and see whether
> it can remain deleted.
>
> I hate bloated  'just in case' include and define blocks. They
> consume your time when trying to trace a bug.

How does having redundant includes affect tracing?