Re: Emacs for Mac OS X 25.1 crashes at start under macOS 10.12.4 beta 1

[Artemio González López]

On 4 Feb 2017, at 10:02 PM, Anders Lindgren <address@hidden> wrote:

To pinpoint what goes wrong in an NS Emacs, you can edit "nsterm.h" and uncomment the line defining the NSTRACE_ENABLED macro. If you run Emacs from a terminal window, you will get debug output of all relevant functions called, hopefully this will tell you what went wrong.

    — Anders

On 4 Feb 2017, at 8:50 PM, Eli Zaretskii <address@hidden> wrote:

Do you have any idea where does this message come from?  There's no
such text in Emacs sources, AFAICS.

On Sat, Feb 4, 2017 at 7:54 PM, Artemio González López <address@hidden> wrote:

Emacs 25.1 (as well as 25.1.1rc) crashes at start under the latest developer beta of macOS (10.12.4 beta 1), producing the error message

2017-01-25 23:50:19.458 Emacs-x86_64-10_9[864:21038] Error -
planarGetBytes count (65536) is not a multiple of
bytesPerRow (96).

I've already reported this as a possible bug, but would like to check if anybody has a pointer as to what maybe the problem and how to fix it.  Any help would be appreciated.

Thank you all very much! Anders, I am now away from home, but when I come back (probably Monday) will try to recompile emacs with the NSTRACE_ENABLED macro enable and post the debug output. In the meantime, I am attaching the macOS full crash log, in case anybody can make something of it. Eli, I checked the emacs source code and found these two instances of bytesPerRow in src/nsimage.m (no trace of the other variable, planarGetBytes, though):

tillina:~/Downloads/emacs artemio$ grep -C5 --colour=always -n bytesPerRow -r .

./src/nsimage.m-215-  bmRep = [[NSBitmapImageRep alloc] initWithBitmapDataPlanes: NULL

./src/nsimage.m-216-                                    pixelsWide: w pixelsHigh: h

./src/nsimage.m-217-                                    bitsPerSample: 8 samplesPerPixel: 4

./src/nsimage.m-218-                                    hasAlpha: YES isPlanar: YES

./src/nsimage.m-219-                                    colorSpaceName: NSCalibratedRGBColorSpace

./src/nsimage.m:220:                                    bytesPerRow: w bitsPerPixel: 0];

./src/nsimage.m-221-

./src/nsimage.m-222-  [bmRep getBitmapDataPlanes: planes];

./src/nsimage.m-223-

./src/nsimage.m-224-  if (fg == 0 && bg == 0)

./src/nsimage.m-225-    {

--

--

./src/nsimage.m-327-                                  pixelsWide: width pixelsHigh: height

./src/nsimage.m-328-                                  /* keep things simple for now */

./src/nsimage.m-329-                                  bitsPerSample: 8 samplesPerPixel: 4 /*RGB+A*/

./src/nsimage.m-330-                                  hasAlpha: YES isPlanar: YES

./src/nsimage.m-331-                                  colorSpaceName: NSCalibratedRGBColorSpace

./src/nsimage.m:332:                                  bytesPerRow: width bitsPerPixel: 0];

./src/nsimage.m-333-

./src/nsimage.m-334-  [bmRep getBitmapDataPlanes: pixmapData];

./src/nsimage.m-335-  for (i =0; i<4; i++)

./src/nsimage.m-336-    memset (pixmapData[i], 0, width*height);

./src/nsimage.m-337-  [self addRepresentation: bmRep];

By the way, bytesPerRow seems to be a variable used by Apple’s Swift language that returns the number of bytes allocated for a single row of a bitmap image (see https://developer.apple.com/reference/coregraphics/cgimage/1455425-bytesperrow).

Thanks again,

Artemio

Emacs-x86_64-10_9_2017-01-30-110350_tillina.crash
Description: Binary data

Artemio Gonzalez Lopez
address@hidden