Re: Problem with accents on freebsd but not on windows

At 11:49 18/04/2015 +0000, Ian Lord wrote:


>I am using Ghostscript 9.15 built from ports on FreeBSD 
>(/usr/ports/print/ghostscript9-agpl-nox11/)
>I also compiled the port (/usr/ports/print/gsfonts/)
>
>So I am "building from sources" technically, but I don't know if the 
>person that built the port changed a lot in it or not...

That's not 'built from source' as I meant it, I mean 'build from the source 
we distribute' and in particular use the 3rd party libraries we supply, not 
any other versions you happen to have.


>I am a bit confused about the bitmap with white text on it, seems like 
>text to me from what I can see (but I'm really not sure). William seems to 
>say the same thing.

There is no white text, there is *invisible* text. Text in PDF can one of 
have several rendering modes; stroke, fill, stroke and fill, neither stroke 
nor fill (and some others).

In the case where the accents are present, what you see rendered is a 
bitmap. On top of that bitmap is invisible (neither stroked nor filled) 
text, which is what is selected when you use a text selection tool on that 
portion of the PDF.

Just because it *looks* like text doesn't mean it is.

So the areas which appear to be correct are correct only because the font 
isn't actually drawn. Where the font is drawn, there are no accented glyphs.


>Let say it's a bitmap with white text over it. What could cause the 
>"accents" to be removed in this case ?

That's not a terribly accurate definition. The accents aren't removed. 
Instead the accented glyphs (not just the accents, but the total glyph) are 
not drawn


>  Ghostscript 9.15 on FreeBSD removes them, but not ghostscript 9.15 on 
> windows ?

That would be (I guess) because you are using different source code on 
Windows and FreeBSD. Before we can realistically even discuss this, you 
need to be using identical ode bases on each platform. THat means you have 
to build from our source, not some packaged version which may include 
random patches, build different 3rd party libraries or pretty much anything 
else you can think of that might break our code.


>  I'm really confused on what would be the cause, beside a bug in 
> ghostscript, if it's not a font substitution problem

A broken set of source would be my guess. Again, retrieve a copy of our 
code, build from that, and test again. Since the font being used is 
embedded in the file its almost inconceivable that it could be a font 
substitution problem. If it is a font substitution, then that's probably 
caused by some change to the source as well.

As I stated before gs-devel is not a support forum, please use one of the 
methods I mentioned in my previous mail. IRC is probably your best bet as 
we will close a bug report which doesn't use our source code, because we 
are not in a position to debug that.

                 Ken