88282 - Xlib-toolkit uses XCreateGC()/XDrawString16()/XFreeGC() instead of using the GC-cache

Status: VERIFIED FIXED

(Core :: XUL, enhancement)

Description

[Roland Mainz]

Xlib-toolkit uses a very inefficient way to draw some fonts:
(from mozilla/gfx/src/xlib/nsFontMetricsXlib.cpp)
-- snip --
  {
    /* XXX is this the right way to do it? Can I get GCCache to give me a
     * new GC? */
    GC copyGC;
    XGCValues values;
    memset(&values, 0, sizeof(XGCValues));

    XGetGCValues(aSurface->GetDisplay(), *gc, GCForeground | GCBackground,
        &values);
    values.font = mFont->fid;
    copyGC = XCreateGC(aSurface->GetDisplay(), aSurface->GetDrawable(),
        GCForeground | GCBackground | GCFont, &values);

    /* note the length must be divided by 2 for X*16 functions */
    XDrawString16(aSurface->GetDisplay(),
                  aSurface->GetDrawable(),
                  copyGC,
                  aX, aY + mBaselineAdjust, (XChar2b *)p, len / 2);
    XFreeGC(aSurface->GetDisplay(), copyGC);
    gc->Release();
    textWidth = XTextWidth16(mFont, (XChar2b *)p, len / 2);
  }
-- snip --

It would be far more efficient if we could use the GC-cache instead of fireing a
XCreateGC()/XFreeGC() sequence for drawing each single string...

Suggested fix:
- Create the following API or make it accessible to nsFontMetricsXlib class:
  - nsRenderingContextXlib::UpdateGC()
  - nsRenderingContextXlib::GetCurrentFont()  
  - nsRenderingContextXlib::SetCurrentFont()
- Use this new API to get rid of this BAD code...

Comment 1

[Roland Mainz]

Accepting. Looks easy... :-)

Comment 2

[Roland Mainz]

Attachment: patch for 2001-06-27-08-trunk

Comment 3

[Roland Mainz]

Filed patch for both Xlib-toolkit and Xprint.

Requesting r= from pocemit ...

Comment 4

[pocemit]

The patch looks great, and works great.
r=pocemit

Looking for sr=

This patch makes a nice performance improvement on a page with lots of mixed
doublebyte/normal fonts.

Comment 5

[tor]

sr=tor

Comment 6

[Roland Mainz]

CC:'ing [e-mail] for checkin, please...

Comment 7

[Mike Kaply [:mkaply]]

Fix checked in.

Comment 8

[pocemit]

massive speed improvement noted, verifying.