Update:  another patch with 1) suggested changes, 2) delete old html before 
generating new, and 3) added flex names for the more complex regular 
expressions.

Exercising the “ninja doxygen” command, brought up some issues.

  1.  The generated html directory contained old as well as new pages.
The build script now deletes old pages before building new ones.
  2.  Using a line of dashes to suppress text formatting is not implemented.
Some dashed comments are harder to read, especially around code samples.
  3.  Comments at the end of “#define” are considered part of the define
and not annotated as doxygen comments.

The first issue was very confusing, so it has been fixed.
My preference is to make the other two issues part of a future enhancement. 
Thoughts?

I’ve also been scanning the generated pages looking for anomalies.

  1.  About 1/3 of pages examined.
  2.  In most cases, the filter works as expected.
  3.  Many places are missing information, so the output has blank fields (as 
expected).
  4.  A few places have obvious nonsense. (eg. a group comment applied to one 
element)
  5.  No situations where the output would be misleading.
  6.  In all cases, the source code is “a click away”.

While I had planned to look at *every* page, I’ll probably stop at the 1/3 
sample – unless someone wants to help scan through the pages with me.

I also heard back from Jetbrains about incorporating custom Doxyfiles. They do 
their own rendering and do not invoke the doxygen command. Custom Doxyfiles are 
not going to happen. (It’s probably the same for VS.)

On my Mac M1, generating doxygen takes about 20 seconds. When graphs are added, 
it takes 5 minutes.


Attachment: doxygen_v5.patch
Description: doxygen_v5.patch

Reply via email to