WebCGM 2.0 "static10" Implementation Matrix
15 September 2006

Latest version:

http://.../static10-matrix.html

This version:

http://.../s10matrix-2006mmdd.html

Previous version:

http://.../s10matrix-2006mmdd.html

Introduction

This document describes the results of WebCGM 2.0 implementations tested against those static-graphics tests of the WebCGM 2.0 Test Suite that are derived from and mostly identical to the static tests of the WebCGM 1.0 Test Suite (whose ZIP file can be downloaded directly.)

Information about the implementations and a legend explaining the table are found after the table.

WebCGM 1.0/2.0 static tests

Implementations Tested

The following table lists the implementations that are covered by the test results.

** Nickname used for the viewer in the below tables (to distinguish multiple viewers from a Company).

Legend

*** 'indef' means that vendor cannot estimate or commit, when it will pass.

Comments from table

• af1 -- a problem with unambiguous interpretation of ISO CGM:1999 D.4.5.12 has been raised. Withdrawal of this test is being considered until there is a clarifying ruling or erratum about CGM:1999 D.4.5.12.
• iv1 -- this test and several others show variation in pattern size. However that is not part of the verdict criteria of the test, so it's status is "Ok" (Pass).
• iv2 -- for PATTBL02, it is clear that the implementation uses the correct patterns, but the size is off. "Partial" is probably the best status, as the size is related to the verdict criteria, if only indirectly. PATTBL01 is similar, except the reference PNG is too-low resolution to discern the exact pattern(s).
• pc1 -- the test admits to variation in exact rendering of dashes, hence where corners fall in dash pattern, hence presence/absence of mitres on some corners.
• pt1 -- this test is inappropriate for typical screen display viewers, because of extreme resolution demands relative to typical (Windows) screen resolutions. It should therefore not be counted, or alternatively, any viewer should pass as long as it shows different stuff within the boxes. (The test could be good for a device like a PostScript printer.) Also the reference PNG is useless because it has less than 1/2 pixel per pattern-definition cell.
• ti1 -- Partial pass, the two upper left arcs are missing.
• ti2 -- Partial pass, cell array is not the correct size.
• ti3 -- Partial pass, the pattern interiors are not correctly rendered.
• ti4 -- Partial pass, the two upper left arcs are missing.
• ti5 -- Note. The PNG image is incorrect. In the upper right of the picture there are Disjoint Polylines with line types 2, 3, 4, and 5 (two of the latter). None of these lines are correctly rendered. The type 3 (dot) line appears as a type 2 (dash) line. The other line types 2, 4, and 5 all appear to be the same type and most closely resemble registered type 10 (chain). We fail because we do not handle the degeneracy at 350, 200.
• ti6 -- Partial pass, the pattern interiors are not correctly rendered.
• ti7 -- Partial pass, butt OK, other two fail.
• ti8 -- Partial pass. The PNG is incorrect. Line type -7 has 0 for the dash lengths in the LETD element. According to CGM99, these are supposed to be drawn as dots. A line of this type occurs in the CGM file beginning at (700,500) and nothing is visible at that location in the PNG file. Also line type -4 has a dot in its definition, and a line of this type occurs beginning at (900,800). Again, the dot is not apparent in the PNG file.
• ti9 -- Partial pass, mitre OK, other two fail.