[gs-cvs] rev 7486 - trunk/gs/doc
henrys at ghostscript.com
henrys at ghostscript.com
Sun Dec 10 11:50:25 PST 2006
Author: henrys
Date: 2006-12-10 11:50:24 -0800 (Sun, 10 Dec 2006)
New Revision: 7486
Modified:
trunk/gs/doc/pscet_status.txt
Log:
Updates from Peter.
Modified: trunk/gs/doc/pscet_status.txt
===================================================================
--- trunk/gs/doc/pscet_status.txt 2006-12-10 09:52:06 UTC (rev 7485)
+++ trunk/gs/doc/pscet_status.txt 2006-12-10 19:50:24 UTC (rev 7486)
@@ -1185,7 +1185,7 @@
09-49D-2 OK Fixed by r7123. Color corrected by rev 7172.
-09-49D-3 DIFF GS gets '1' status, MEMO 3 & 4. assign: Peter.
+09-49D-3 OK Fixed by r7450
09-50-1 DIFF GS has different line spacing. CPSI and Tek agree. GS gets same
results for "false setstrokeadjust" "true setstrokeadjust".
@@ -1260,7 +1260,10 @@
09-55-2 OK
09-56-1 DIFF Device Dependent: Tek and Adobe show '/findcolorrendering' instead
- of '--findcolorrendering--' with errors. Assign: Peter
+ of '--findcolorrendering--' with errors.I've checked in a fix for
+ the setcolorrendering issue. There is still a discrepancy in that
+ the reference output shows additional built-in color rendering names
+ /GenericImagesetter and /AppleMonitor (peter). Assign ray
09-56-2 OK revisin 7234 (peter)
@@ -1474,7 +1477,17 @@
10-15-2 DIFF DIfferences in the column labeled 'MUCH'. CPSI and Distiller/
Acrobat render a '1' in item 22. GS does not. GS renders a '1'
- in item 24. CPSI and Distiller/Acrobat do not. Assign: Peter
+ in item 24. CPSI and Distiller/Acrobat do not.
+ -
+ The diffs in items 'much22' and 'much24' are graphics-related (user paths /
+ insideness testing), not interpreter-related. Please assign them to someone
+ else.
+ -
+ The diff in item 'gray05' (reference = blank, gs = 4)
+ does appear to be interpreter-related (the same diff
+ occurs in 13-24-6): I'll continue to work on
+ this. (peter) Assign (much 22 and 24): Igor
+ (other issues fixed with r7452 Peter).
10-16-1 OK
@@ -2746,9 +2759,9 @@
13-01-1 OK Minor differences visually reviewed by RJJ
-13-01-2 DIFF in cpsi the ashow output is produced before the
- rangecheck. Also gs is missing the 'A' to the right
- of the first rangecheck problem.
+13-01-2 DIFF cpsi keeps the operands on the stack for the last 4 errors (one
+ nocurrentpoint, 3 rangecheck), gs does not. Also, gs displays
+ an 'A' for the first rangecheck error, cpsi does not.
assign Peter.
13-01-3 OK Minor differences visually reviewed by RJJ
@@ -4008,8 +4021,12 @@
23-07-1 OK Fixed by rev. 7442
-23-07-2 DIFF GS shows command error as '--defineresource--' and '--findresource--'
- Tek and CPSI return 'defineresource' and 'findresource'. assign: Peter
+23-07-2 AOK There is still a diff in 23-07-2 in that the contents of our resource
+ category dictionaries (in this case, the Generic dictionary) are different
+ from the Adobe interpreters: we and Adobe have different implementation-specific
+ (non-PLRM) entries, and the PLRM-documented procedures (FindResource, etc.)
+ are different. This is inherently implementation-dependent and does not
+ affect any real program, so no further changes are needed. (Peter)
23-07-3 OK Fixed by rev. 7442
@@ -4017,9 +4034,21 @@
23-08-1 OK
-23-09-1 DIFF GS shows -1 for bytesavailable (Device Dependent). GS get errors
- 'undefinedfilename' and 'ioerror' vs. CSPI and Tek 'invalidaccess',
- and in other places 'undefinedfilename'. assign: Peter
+23-09-1 DIFF The first bytesavailable diff is identified as "highly device dependent".
+ As such, I recommend discounting it. No real program will rely on the
+ result.
+ -
+ The two invalidaccess diffs occur because the reference implementation does
+ not allow PostScript programs to open files for writing. This is not a gs
+ bug, but it does suggest that gs should provide some way to indicate/detect
+ that it is running in an embedded environment with no (writable) filesystem.
+ Does this exist already?
+ -
+ The remaining 3 diffs occur because gs implements %null as a special file
+ name and the reference implementation apparently doesn't, per a previous
+ e-mail.
+ -
+ assign: ray
23-10-1 AOK Device Dependent: GS looks like it is working similar to CPSI ot
Tek, but all three differ. RJJ
@@ -4197,12 +4226,9 @@
23-17-1 OK Minor differences visually reviewed by RJJ
-23-17-2 DIFF GS doesn't get errors on some ILLEGAL tests -- (null 0 printobject flush)
- and after. Tek and CPSI return --printobject-- instead of --writeobject--
- or --.writeobjects--. GS shows 'undefined' instead of 'typecheck' on
- several lines. Assign: Ray
+23-17-2 AOK Probably fixed with 23-17-3
-23-17-3 DIFF GS has blank for MORE 27, CPSI and Tek return '2'. Assign: Peter
+23-17-3 AOK Fixed see 23-29-2, 23-36-3 and 23-36-4
23-18-1 OK Minor differences visually reviewed by RJJ
@@ -4270,7 +4296,7 @@
23-29-1 OK
-23-29-2 DIFF GS missing error from printobject when objectformat == 0: Assign Peter
+23-29-2 OK fixed in rev. 7232 (Peter)
23-29-3 OK
@@ -4326,11 +4352,9 @@
This causes a the end of the string to shift to the right on lines that
mismatch. Assign: Igor.
-23-36-3 DIFF GS not detecting errors when objectformat == 0. Similar to 23-29-2 with
- writeobject. Assign: Peter
+23-36-3 AOK fixed in rev. 7232 (Peter)
-23-36-4 DIFF more 27: problem with "0 setobjectformat" not detected by GS. Tek and
- CPSI agree and print '2' (similar to 23-17-3). Assign: Peter
+23-36-4 OK fixed in rev. 7232 (Peter)
23-37-1 OK
@@ -4455,8 +4479,8 @@
25-10-2 OK
-25-11-1 DIFF GS gets 'typecheck' instead of 'stackunderflow' from ( log )
- Igor assigns this to Peter.
+25-11-1 OK GS matches Tek hardcopy. Fixed by rev 7140.
+ GS has more accurate (and correct) sin/cos functions than CPSI
25-11-2 OK
@@ -4585,9 +4609,20 @@
27-03-1 DIFF Error messages differ. Ghostscript gives "undefinedfilename"
CPSI gives "undefined".
-27-03-2 DIFF CPSI has ten '1' instances, but GS has none.
- filt18 is an invocation of filenameforall.
- Analyzed by Igor, then Peter.
+27-03-2 AOK test unreasonable:
+ This page enumerates all the IODevices, arbitrarily selects the last one,
+ and then runs the GIF test, checking whether the currentdevparams values for
+ that IODevice change as a result of running the test procedures. This is
+ dependent both on the set of IODevices (which is implementation-dependent,
+ although gs could be expected to include at least the non-hardware-dependent
+ ones documented in the PLRM3, which I believe it does) and the order in
+ which they are enumerated by /IODevice resourceforall (which is totally
+ implementation-dependent).
+ -
+ It appears that in the reference output, the last-enumerated IODevice is one
+ whose devparams values change somewhat unpredictably -- probably %Calendar%.
+ In the gs output, it's one whose devparams values don't change -- i.e. any
+ of the others! (peter)
27-04-1 DIFF Installation dependent (FontResourceDir).
Inconforming currentsystemparams. Analyzed by Igor.
@@ -5032,8 +5067,7 @@
30-04-3 OK
-30-04-4 DIFF currentcolorspace output is [:/DeviceGray:] (packed) in gs.
- assign Peter
+30-04-4 OK fixed in packed array colorspace fix (ray) don't recall the rev.
30-04-5 OK
More information about the gs-cvs
mailing list