Inconsistent framebuf format

[AncientJames]

I'm not sure if this is the place to flag it, but the framebuf FRAMEBUF_GS2_HMSB format is arranged with msb to the right, the reverse of the other formats.

[peterhinch]

Are you sure? Looking at the sourcecode it's hard to see how this can be the case.

[AncientJames]

Sorry, it's GS4 that's the wrong way round.

[peterhinch]

I suggest you back up your assertion by pointing to an error in the very simple getpixel source. As far as I can see there is no possible mechanism for bit reversal.

[AncientJames]

GS2 shifts right with increasing x. GS4 tests bit 0 and then returns the right shifted result if it's unset - ie with decreasing x.

This shows the behaviour:

import framebuf

def fbtest(fmt):
    ba = bytearray(8)
    fb = framebuf.FrameBuffer(ba, 8, 1, fmt)
    fb.fill(0)
    fb.pixel(0, 0, 255)
    return ba[0]

print(f'{fbtest(framebuf.MONO_HLSB):08b}  MONO_HLSB')
print(f'{fbtest(framebuf.MONO_HMSB):08b}  MONO_HMSB')
print(f'{fbtest(framebuf.GS2_HMSB):08b}  GS2_HMSB')
print(f'{fbtest(framebuf.GS4_HMSB):08b}  GS4_HMSB')

Resulting in:

10000000  MONO_HLSB
00000001  MONO_HMSB
00000011  GS2_HMSB
11110000  GS4_HMSB

[peterhinch]

OK, now I understand.

I think we need new constants with a different naming convention and much better documentation. We can't delete the existing constants because it would break code. I suggest GS2_RJ == GS2_HMSB and GS4_LJ == GS4_HMSB where LJ and RJ mean left and right justification for multi-bit colors.

Do you want to raise a PR or shall I?

Any other thoughts on naming?