Hello,
Just wanted to follow up on this. We found the bug (a single missing character
in a regex). A PR has been opened on github
https://github.com/doxygen/doxygen/pull/7192
Cheers,
Ryan
On Thu, 31 Jan 2019, Ryan S. Elliott wrote:
Hi Travis,
Thanks for your reply! This all sounds reasonable. I won't be able to
dig-in to this right away, but I'll put it on my list to get back to soon.
Thanks,
Ryan
On Thu, 31 Jan 2019, Travis Everett wrote:
Ryan,
I don't know a ton about how those parts of the codebase work so I stalled
a bit hoping someone else would respond (there's some chance I'll mis-lead
you here...). In any case, a good place to start is by seeing whether the
behavior changed in 1.8.12/13/14. That should let you narrow your focus to
just commits in a single version.
If you aren't set up to build Doxygen from source (and/or aren't
comfortable doing so), it's probably fine to open an issue report with the
behavior you've seen, which version changed it, and a minimal set of
files/config necessary to reproduce it. If you are comfortable building
from source you can push this further and identify the specific commit.
Hopefully the issue you're seeing was caused by an update to
src/fortranscanner.i or src/fortrancode.i (there are probably only a few
commits touching those per version). You can use git or github to identify
which commits between two versions touched either of those files, and
recompile/rebuild (depending on how long your docs take to build you may be
able to speed this up a lot by making a separate config that just includes
the minimum set of files you need to see the behavior) at each to see which
introduces the break.
If that doesn't work (or if there are more commits than I suspect touching
those two files) you can use the more general git bisect command (along
with a recompile and rebuild doc cycle) to narrow it down to the last-good
commit.
An issue report that identifies the commit/PR introducing the error should
make it a quicker fix; if you're thinking of tackling the fix itself but
need help figuring out where to start, it may help to fish around on the
separate development list
HTH,
Travis
On Wed, Jan 30, 2019 at 5:51 PM Ryan S. Elliott <relli...@umn.edu> wrote:
Hello,
Thanks for the reply.
I tried 1.8.11 on my mac and I get a Segmentation fault!
.
.
.
Reading
/Users/relliott/unison-sync/KIM/git/kim-api/fortran/include/kim_model_compute_arguments_module.f90...
Parsing file
/Users/relliott/unison-sync/KIM/git/kim-api/fortran/include/kim_model_compute_arguments_module.f90...
/bin/sh: line 1: 54543 Segmentation fault: 11
/Users/relliott/doxygen/build/bin/doxygen
/Users/relliott/unison-sync/KIM/git/kim-api/build/docs/Doxyfile.docs
Anyway, then I set up a ubuntu VM and got it to work. Indeed, 1.8.11 on
ubuntu
it seems I get the same output as RTD. I've also confirmed that 1.8.14 on
ubuntu produces the same output (without the interfaces documented) that I
previously obtained on my mac.
So, it seems that this is a regression in behavior from 1.8.11 to 1.8.14?
How
should I go about figuring out when/where this happened and what to do
about
it?
On Wed, 30 Jan 2019, Travis Everett wrote:
The copy on RTD indicates it was generated by doxygen 1.8.11, while the
one
at openkim indicates 1.8.14; a good first step would be installing 1.8.11
and seeing if your output matches what's on RTD.
On Wed, Jan 30, 2019 at 3:08 PM Ryan S. Elliott <relli...@umn.edu>
wrote:
Hello,
I'm working on the doxygen documentation for the kim-api project
(https://github.com/openkim/kim-api)
When I locally (on my mac) generate the doxygen docs I see problems with
the
fortran docs. In particular, there are a fair number of functions that
are not
automatically documented by the parser. Further, no generic interfaces
are
documented.
However, the exact same git commit, when used to generate the doxygen
docs
by
readthedocs documents all the functions and generates the generic
interface
docs.
Here is a link to a readthedocs page showing the interfaces:
https://kim-api.readthedocs.io/en/latest/namespacekim__model__module.html
Here is a link to a similar page generated locally:
https://openkim.org/kim-api/docs-beta.3/namespacekim__model__module.html
In particular, also notice that in the locally generated docs the
"kim_model_is_routine_present" function is not documented!
I can't understand why these differences are occuring. Can anyone point
me in
the right direction?
Thanks,
Ryan Elliott
_______________________________________________
Doxygen-users mailing list
Doxygen-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/doxygen-users
--
Ryan S. Elliott, Ph.D. and Professor
Aerospace Engineering & Mechanics, University of Minnesota
(612) 624-2376 (626-1558 fax)
https://z.umn.edu/relliott
download vCard <https://z.umn.edu/relliott_vcf>
----------
The distinction between past, present and future, is only an illusion,
even
if a stubborn one.
Albert
Einstein
----------
--
Ryan S. Elliott, Ph.D. and Professor
Aerospace Engineering & Mechanics, University of Minnesota
(612) 624-2376 (626-1558 fax)
https://z.umn.edu/relliott
download vCard <https://z.umn.edu/relliott_vcf>
----------
Words ought to be a little wild for they are the assaults of thought on the
unthinking.
John Maynard Keynes
----------
_______________________________________________
Doxygen-users mailing list
Doxygen-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/doxygen-users
