Possible bug in Pro text API

[cormullion]

In some Luxor testing I found a problem with the Pro text API, and here's the same problem reproduced in Cairo:

The blue area has been applied to the second "Hamburgefons" string, looks like it's inherited from the previous text with markup. I'm hoping this is a bug -- but of course I may have misunderstood how it's supposed to work.

## header to provide surface and context
using Cairo
c = CairoRGBSurface(256,256)
cr = CairoContext(c)

Cairo.save(cr)
set_source_rgb(cr,0.8,0.8,0.8)    # light gray
rectangle(cr,0.0,0.0,256.0,256.0) # background
Cairo.fill(cr)
restore(cr)

Cairo.save(cr)

set_font_face(cr, "Sans 16")

Cairo.text(cr,16.0,40.0,"Hamburgefons")
Cairo.text(cr,16.0,72.0,"sp⁰¹²³,min⁻²,αΑβΒφϕΦγΓ")
Cairo.text(cr,16.0,104.0,"Text<b>Bold</b><i>Italic</i><sup>super-2</sup>",markup=true)
Cairo.text(cr,40.0,224.0,"Es geht <span foreground=\"white\" background=\"blue\">aufwärts</span> !",markup=true,angle=30.0)

# But this doesn't have markup...:

Cairo.text(cr,16.0,223.0,"Hamburgefons again")

#using textwidth and height 

set_font_face(cr, "Sans 12")

a = "A"
aheight = textheight(cr,a)
awidth = textwidth(cr,a)
atext = @sprintf("%s wd=%2.2f,ht=%2.2f",a,awidth,aheight)
Cairo.text(cr,16.0,240.0,atext,markup=true)


## mark picture with current date
Cairo.restore(cr)
move_to(cr,0.0,12.0)
set_source_rgb(cr, 0,0,0)
show_text(cr,Libc.strftime(time()))
write_to_png(c,"/tmp/sample_pango_text.png")

You can also get other aspects of the markup leaking into other strings (eg bold, superscript).

julia> Pkg.status("Cairo")
 - Cairo                         0.3.1

julia> versioninfo()
Julia Version 0.6.0
Commit 903644385b (2017-06-19 13:05 UTC)
Platform Info:
  OS: macOS (x86_64-apple-darwin13.4.0)
  CPU: Intel(R) Core(TM) i5-2500S CPU @ 2.70GHz
  WORD_SIZE: 64
  BLAS: libopenblas (USE64BITINT DYNAMIC_ARCH NO_AFFINITY Sandybridge)
  LAPACK: libopenblas64_
  LIBM: libopenlibm
  LLVM: libLLVM-3.9.1 (ORCJIT, sandybridge)

[lobingera]

I have recognized similar problems when writing the demo code. I'm not a pango expert, but i think the main problem is reusing the pango-context in multiple lines. You could try to create a new Cairo.context which creates a new pango context implicitly.

In the long-run, a new text API and doing a clean split of cairo and pango seems to be needed,

[lobingera]

#226