Re: [Doxygen-users] how to avoid running dot for really big diagrams

[Oliver S. <oli...@ut...>]

----- Original Message -----
From: "Dimitri van Heesch" <di...@st...>
To: <dox...@li...>
Sent: Saturday, March 29, 2003 9:09 AM
Subject: Re: [Doxygen-users] how to avoid running dot for really big
diagrams


> within the specific image size. Doxygen uses a binary search, starting
> with the smallest graph (depth from root=1) and the complete graph
> (depth from root=N). When the complete graph doesn't fit, one with maximum
> depth of N/2 is tried, etc. This obviously requires multiple runs of dot
> (and at least one with the complete graph) to determine the best fit.
>
> So if you stop dot when the number of nodes exceed a certain value that
> will give better results. Maybe a better way to reduce time is to add an
> config option to limit the graph depth (i.e. distance from the root) or
> maximum node count, then that could be used as the upper bound for
searching
> a best fit. Let me know which one (graph depth or node count) you think is
> better.

Dimitri, the above is a great idea and is something that I've been wishing
for a while. I have found over time that in the docs, the nodes at depths
more than 3 or 4 from root are rarely relevant to understanding the root
class, and just make the graph more complicated than necessary (i.e. deters
users from using the info). Also, judging by the original posters tests,
such a config option would probably greatly accelerate the run of doxygen on
complex systems. Verifying this would be easy: make the loop start at the
min(N, DOT_MAX_DEPTH). If you (or someone) tell me where that loop is I
could do a test of this on my end and I can let you know how it goes.

Oliver