bug#37752: Scaling factor inconsistencies between X with and without GTK

bug#37752: Scaling factor inconsistencies between X with and without GTK

[Carlos Pita]

In many places in xterm.c you have code like:

#ifdef USE_GTK
  scale = xg_get_scale (f);
#endif

Sometimes scale is only used in GTK specific code and that's ok.

But sometimes it's used in code that's not particular to GTK, for example:

int scale = 1;
#ifdef USE_GTK
  scale = xg_get_scale (f);
#endif
FRAME_CR_SURFACE (f) =
    cairo_image_surface_create (CAIRO_FORMAT_ARGB32,
      scale * FRAME_PIXEL_WIDTH (f),
      scale * FRAME_PIXEL_HEIGHT (f));

Besides, there is x_get_scale_factor, also defined in xterm.c, which
computes the scaling factor in a GTK-independent way, and also
discriminates between x and y scaling factors. It's only used by
x_draw_underwave.

I would like to abstract both cases inside x_get_scale_factor, that
would then use xg_get_scale if in GTK or compute just one scaling
factor (for x or for y) otherwise. Notice that given that the scaling
factor is the ratio of the effective resolution to a base resolution
of 96 dpi *truncated down* to the nearest integer, AFAICS it's almost
impossible to get different scaling factors for x and for y.

What do you think?

Best regards
--
Carlos

bug#37752: Scaling factor inconsistencies between X with and without GTK

[Robert Pluim]

>>>>> On Mon, 14 Oct 2019 21:21:54 -0300, Carlos Pita <carlosjosepita@gmail.com> said:

    Carlos> In many places in xterm.c you have code like:
    Carlos> #ifdef USE_GTK
    Carlos>   scale = xg_get_scale (f);
    Carlos> #endif

    Carlos> Sometimes scale is only used in GTK specific code and that's ok.

    Carlos> But sometimes it's used in code that's not particular to GTK, for example:

    Carlos> int scale = 1;
    Carlos> #ifdef USE_GTK
    Carlos>   scale = xg_get_scale (f);
    Carlos> #endif
    Carlos> FRAME_CR_SURFACE (f) =
    Carlos>     cairo_image_surface_create (CAIRO_FORMAT_ARGB32,
    Carlos>       scale * FRAME_PIXEL_WIDTH (f),
    Carlos>       scale * FRAME_PIXEL_HEIGHT (f));

I canʼt find that code in master.

    Carlos> Besides, there is x_get_scale_factor, also defined in xterm.c, which
    Carlos> computes the scaling factor in a GTK-independent way, and also
    Carlos> discriminates between x and y scaling factors. It's only used by
    Carlos> x_draw_underwave.

    Carlos> I would like to abstract both cases inside x_get_scale_factor, that
    Carlos> would then use xg_get_scale if in GTK or compute just one scaling
    Carlos> factor (for x or for y) otherwise. Notice that given that the scaling
    Carlos> factor is the ratio of the effective resolution to a base resolution
    Carlos> of 96 dpi *truncated down* to the nearest integer, AFAICS it's almost
    Carlos> impossible to get different scaling factors for x and for y.

Sure, that makes sense. Do you have a test case showing a display
difference, or is this just code cleanup?

Robert

bug#37752: Scaling factor inconsistencies between X with and without GTK

[Carlos Pita]

>     Carlos> int scale = 1;
>     Carlos> #ifdef USE_GTK
>     Carlos>   scale = xg_get_scale (f);
>     Carlos> #endif
>     Carlos> FRAME_CR_SURFACE (f) =
>     Carlos>     cairo_image_surface_create (CAIRO_FORMAT_ARGB32,
>     Carlos>       scale * FRAME_PIXEL_WIDTH (f),
>     Carlos>       scale * FRAME_PIXEL_HEIGHT (f));
>
> I canʼt find that code in master.

You're right, this and other xg_get_scale usages were removed wrt
emacs-26. Just one of them remains.

> Sure, that makes sense. Do you have a test case showing a display
> difference, or is this just code cleanup?

It's just code cleanup.

Best regards
--
Carlos

bug#37752: Scaling factor inconsistencies between X with and without GTK

[Robert Pluim]

>>>>> On Tue, 15 Oct 2019 05:09:04 -0300, Carlos Pita <carlosjosepita@gmail.com> said:

    Carlos> int scale = 1;
    Carlos> #ifdef USE_GTK
    Carlos> scale = xg_get_scale (f);
    Carlos> #endif
    Carlos> FRAME_CR_SURFACE (f) =
    Carlos> cairo_image_surface_create (CAIRO_FORMAT_ARGB32,
    Carlos> scale * FRAME_PIXEL_WIDTH (f),
    Carlos> scale * FRAME_PIXEL_HEIGHT (f));
    >> 
    >> I canʼt find that code in master.

    Carlos> You're right, this and other xg_get_scale usages were removed wrt
    Carlos> emacs-26. Just one of them remains.

If thatʼs in code thatʼs independent of GTK3 usage, then by all means
fix it.

Robert

bug#37752: Scaling factor inconsistencies between X with and without GTK

[Carlos Pita]

Hi Robert, I'm closing this in favor of

#37770 [PATCH] Expose scale factor through the redisplay interface

https://debbugs.gnu.org/cgi/bugreport.cgi?bug=37770

Please take a look at it if you're so kind.

Best regards
--
Carlos