Re: Internal Error (overlap) for some fonts when running make

[James]

On 05/11/2019 12:53, Werner LEMBERG wrote:

> > > Now that Dan has made default make commands 'terser' I am noticing
> > > while building our fonts that I see quite a few 'errors' that look
> > > like this:
> > >
> > > Internal Error (overlap) in clefs.petrucci.c5_change:
> > >    Winding number did not return to 0 when x=25.9951
> > > Internal Error (overlap) in clefs.petrucci.c5_change:
> > >    monotonic is >> both needed and unneeded
> > > (22.1885,310.434)->(25.9951,304.135). x=25.9951 (prev=8.40259)
> > >
> > > Of course they have 'always' (probably) been there - I don't want to
> > > claim anything like a regression.
> It's not a regression, right.

Oh ...

OK when you wrote that I thought I'd check (as much as my skill will let me)

My first probably-not-very-correct thing to do next was check out 
stable/2.18 and make LP from there (on the same system am building 
latest master on) and search for the same errors.

Neither error was posted.

I'd assume we'd still see them *before* Dan's 'make 'make' terse' 
enhancements from a few weeks back and that his changes wouldn't 
suddenly 'add' compilation messages that weren't there before :D

Next I picked some random 2.19 build from staging (that just happened to 
be tagged as 2.19.55-1 Circa ~start of 2017) and built LP from there.

and the errors were there!

Going back a bit more ...

2.19.10-1 Circa ~start of 2014 and built LP from there.

Same errors reported.

Then 2.19.1-1 and built LP from there.

Same errors reported (but 18 less errors than the previous - i.e. 137 vs 
155)

So is this a regression?

> > > But do we need to care?
> We should.
>
> > The first is pretty typical for Metafont/Metapost fonts, affecting
> > cusps.  The latter looks like there might be a logic error in the
> > programming for a variable called "monotonic".  Fixing those kinds
> > of bugs would improve confidence in the font creation instructions,
> > but the real test is whether people complain about the look of the
> > resulting character.
> Usually it is not noticeable.
>
> Note that recently I saw some improvements w.r.t. monotonic stuff in
> FontForge's git repository.  We thus should do two things for testing.
>
> (1) Use a FontForge binary built from current git.

OK

Current installed version (before):

**

$fontforge --version

Copyright (c) 2000-2014 by George Williams. See AUTHORS for Contributors.
 License GPLv3+: GNU GPL version 3 or later 
<http://gnu.org/licenses/gpl.html>
 with many parts BSD <http://fontforge.org/license.html>. Please read 
LICENSE.
 Based on sources from 11:21 UTC 24-Sep-2017-ML-D.
 Based on source from git with hash:
no xdefs_filename!
TESTING: getPixmapDir:/usr/share/fontforge/pixmaps
TESTING: getShareDir:/usr/share/fontforge
TESTING: GResourceProgramDir:/usr/bin
trying default theme:/usr/share/fontforge/pixmaps/resources
fontforge 11:21 UTC 24-Sep-2017
libfontforge 20170924

**

After building from git (you developer's are a funny lot aren't you? 
With all your different build/make tools - I had to 'make breakfast' 
last week (with Bacon and/or Eggs IIRC) building Android,  you'd think 
with the 'foundry' connotation for font's the fony forge developers 
would have invented a build toolset called 'Hammer' or 'Smelt' or 
something (I know, I know!) any way after creating and installing my 
fontforge ...

(with) Ninjas ... :P

I had to copy this exe from usr/loca/share/bin - where the ninja install 
command put it -  in place of the old one in /usr/local/bin (i renamed 
the old one so I can put it back.

I have this now:

***

$ ./fontforge --version
Copyright (c) 2000-2019. See AUTHORS for Contributors.
 License GPLv3+: GNU GPL version 3 or later 
<http://gnu.org/licenses/gpl.html>
 with many parts BSD <http://fontforge.org/license.html>. Please read 
LICENSE.
 Version: 20190801
 Based on sources from 2019-11-05 14:55 UTC-ML-D-GDK3.
 Based on source from git with hash: 
380db18f2ed03c128717adc172f7b8a3a65c926e
Failed to open resource file: 
/home/james/Downloads/git_repos/fontforge/build/share/fontforge/pixmaps/resources
fontforge 20190801
build date: 2019-11-05 14:55 UTC

***

aaand .. making current stable gives me

The same errors

\o/


> (2) Report any conversion errors to the FontForge maintainers.
>      Hopefully, they now have the expertise to fix something (a few
>      years ago, they didn't).

Well before I figure out how to do that (i.e. how to be that random 
internet person posting issues and then getting asked technical 
questions I probably wouldn't know how to answer).

I've attached the list of 'monotonic' errors on their own (there are 
also slightly less 'Winding Number' errors that occur) as well as the 
full make output - it's ~ 150Kb in all. You can grep/search for Winding 
and Monotonic to give you some context.

What do you think?

James

Monotonic errors.txt
Description: Text document

Stdout_make.txt
Description: Text document