129655 - [ps] PostScript module device depth is 1bit

Status: VERIFIED FIXED

(Core :: Printing: Output, defect)

Description

[Roland Mainz]

I found a longstanding issue in the PostScript module:
The device depth is always 1bit instead of 24bit(8bit per R,G,B) - this may
screw-up some things...

Comment 1

[Roland Mainz]

Fix in bug 119491 ...

Comment 2

[Roland Mainz]

See http://lxr.mozilla.org/seamonkey/source/gfx/src/ps/nsDeviceContextPS.cpp#139

Comment 3

[Roland Mainz]

Moved the fix into bug 126799 ...

Comment 4

[dcone (gone)]

I would like to make a couple comments.

1.)  I would like the patch for this bug.. to be placed by itself in this bug,
not as part of another patch.  Its a better way to keep track of this issue, and
what was done to solve this issue.

2.)  If you would like.. I would like to fix this bug, because I dont fully
understand what it means to set the depth to something besides 1 bit deep for
postscript.   Printers are mostly 1 bit deep.. and even if you have a printer
that is not.. I dont know how this parameter would effect that output.  I think
that paremater is more something used for buffers, calculating row distances,
memory, etc, I dont really know how postscript is using that at this moment.

3.)  If you do fix this.. I would require that I review this.. I am the owner of
the postscript stuff and I want to make sure we dont have any unessisary
regressions, or at least I want to keep track of this issues that effect postscript.

4.) What is your take on the depth, and how do you think the PS module should
use it, and what things are you talking about that may get screwed up.

Comment 5

[Roland Mainz]

[e-mail] wrote:
 
> I would like to make a couple comments.
> 
> 1.)  I would like the patch for this bug.. to be placed by itself in this bug,
> not as part of another patch.  Its a better way to keep track of this issue, 

;-((
OK, I'll take the "fix" (=one line) out of bug 126799...

> and what was done to solve this issue.

The fix is to change "depth = 1" to "depth = 24" - that's all.

> 2.)  If you would like.. I would like to fix this bug, because I dont fully
> understand what it means to set the depth to something besides 1 bit deep for
> postscript.   Printers are mostly 1 bit deep.. and even if you have a printer
> that is not.. I dont know how this parameter would effect that output.  I 
> think that paremater is more something used for buffers, calculating row 
> distances, memory, etc, I dont really know how postscript is using that at 
> this moment.

The (future) problem will be SVG - our ScaleableVectorGraphics support.
Currently it simply crashes cause it lacks support for 1bit devices (actually
the PostScript module code itself assumes that it is fed with 24bit RGB data but
annouces itself that it is a 1bit device...)...

> 3.)  If you do fix this.. I would require that I review this.. I am the owner
> of > the postscript stuff and I want to make sure we dont have any unessisary
> regressions, or at least I want to keep track of this issues that effect 
> postscript.

Sure.

> 4.) What is your take on the depth, and how do you think the PS module should
> use it, and what things are you talking about that may get screwed up.

See above. Code which looks at the device depth and adjust itself based on that 
value will go nuts (like our SVG code) ...

Comment 6

[Roland Mainz]

Attachment: Patch for 2002-03-23-08-trunk

Comment 7

[Roland Mainz]

dcone - can you r= the patch, please ?

Comment 8

[Marc Attinasi]

Comment on attachment 76061 [details] [diff] [review]
Patch for 2002-03-23-08-trunk

sr=attinasi (aaahhh, nice and easy)

Comment 9

[dcone (gone)]

are we always 24 bit.. what about on black and white printers.. are we 8 bit 
deep there, since its indexed?

Comment 10

[Roland Mainz]

dcone wrote:
> are we always 24 bit.. what about on black and white printers.. are we 8 bit 
> deep there, since its indexed?

The data returned by |GetDepth()| is the depth we accept, not the depth we
output. nsDeviceContextPS always wants 24bit RGB as input.

Comment 11

[dcone (gone)]

r=dcone.  
I dont think at this time its bad to have 24 bit returned.. but I think its 
important to keep discussing how to use this depth correctly.
The Depth is what is supported by the Device represented by the DeviceContext.  
So for example if a routine asked for the depth of the device, that routine can 
use a depth for a "blend" that is appropriate for that device.  So my question 
.. is 24 bit always appropriate, or are there times it should be 8 bit, or even 
1 bit.  If we are going to output an image that is optimized for a printer, let 
say and 8 bit color indexed printer... I think that 8 bit should be returned so 
a pallete could be optimized so and optimized halftone of that output image 
could be generated.  If we had a super duper color photo printer 24 bit would be 
appropriate.  Or if we had a black and white printer, a depth of 1 bit would be 
important so we could do a halftone thats better than the ordered dither used by 
postscript.  

Comment 12

[dcone (gone)]

Comment on attachment 76061 [details] [diff] [review]
Patch for 2002-03-23-08-trunk

r=dcone

Comment 13

[Roland Mainz]

dcone:
Mozilla does not have any special support for 8bit/1bit displays. For example on
Linux/Unix we use the GDK/GTK+ libraries which only offer a static, hardcoded
palette (which is very worse compared to what NS4.x can do with it's dynamic
palette change code, but those people do not care about palette displays
anyway... ;-( ) and is fed with 24bit RGB data.
Same applies to the PostScript module... the current code always expects 24bit
RGB data - feeding anything else just causes a crash... ;-(

Comment 14

[dcone (gone)]

but the DeviceContext is a virtual device.. so it supports what we say.  If GTK 
supports only a limited number of colors.. it handles that.  Postscript is not 
limited to GTK.  For example.. I might want to output PostScript on a device 
that is a high end printer.. then 24 bit is appropriate.. if I wanted to do 
advanced halftones.. 1 bit or 8 bit is appropriate.... it tells the nsImage that  
we have certain capabilites in this DeviceContext.  I think its a mistake to 
limit how this is used.. and better to use this to reflect the true depth of a 
device.. its representing.

Comment 15

[Roland Mainz]

dcone:
OT: Does PostScript Level 2 (or level 3) support 8bit palette-indexed images ?

Comment 16

[Roland Mainz]

Patch checked in
(http://bonsai.mozilla.org/cvsquery.cgi?module=MozillaTinderboxAll&branch=HEAD&cvsroot=/cvsroot&date=explicit&mindate=1019824680&maxdate=1019825760&who=peterv%25netscape.com),
marking bug as FIXED.

Comment 17

[sujay]

Roland please verify...thanks..

Comment 18

[Roland Mainz]

Verifying, seems to "work" now... :)