[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [List Home]
Subject: Re: [cgmo-webcgm] Review - lineOffset set.
One point of Rob's review is worth a look: can we have a good test of LTIO with generic line types (2,3,4,5), as opposed to types that are precisely defined by LETD? I agree with the point that the results with generic types will vary, and therefore the minimal good test should involve LETD-defined types. Question: since it is legal to have LTIO with a generic type (2,3,4,5), would it be any use to have a separate test-case that exercises that? It would be tricky, because the correct results are sort of generic -- they would have to be described in loose way -- "e.g., the gaps precess to the right for subsequent lines in the group" -- and the reference image would only be one of a large number of correct images, the latter depending on the viewers particular choice of the exact pattern for the generic style (2,3,4,5). -Lofton. At 01:35 PM 2/12/2009 -0700, Lofton Henderson wrote: >It sounds like this one should transition to "reworking" [1]. > >Regards, >-Lofton. > >[1] ftp://ftp.cgmlarson.com/test-matrix.htm > >At 11:16 AM 2/12/2009 -0700, Robert Orosz wrote: >>All, >> >>Below is my review of the lineOffset test. >> >>Regards, >> >>Rob >>Summary >>This test is not acceptable as it stands. There are issues with both the >>HTML markup and WebCGM profile, vide infra, but more importantly, I don't >>think the image as it stands makes a good test. >>Details >>The image consists of six groups of four lines. Each group has four lines >>with the line type and all other line attributes held constant, and the Line >>Type Initial Offset (LTIO) is varied with values of 0.0, 0.25, 0.5, and 0.75 >>in each group. That is a good test. However, of the six groups, four use the >>standard line types 2, 3, 4, and 5. The exact rendering of these line types >>is implementation dependent. Therefore, any offset into the pattern no >>matter how precisely defined, will also be implementation dependent. >>The last two groups use Line and Edge Type defined line types. This is good, >>in fact I think sufficient for this test. I would like to see the image >>scaled a little larger, so it is easier to see. My suggestion would be to >>arrange the lines in each of the two test groups vertically across the >>entire image. Two or three pattern cycles should be sufficient for users to >>get the idea. >>The lines are pretty thick, just over 3.5 points. Looking at the PNG image >>in both Internet Explorer 6 and Internet Explorer 7 gives the impression >>that there are multiple line widths in the drawing, which is not the case. I >>would try much thinner lines, e.g. 00 or 000 points. >>There is no explanation in the image of exactly what the purpose of the red >>lines are. The long one obviously marks the origin of each of the lines in >>the test group, but there is no explanation of the purpose of the shorter >>lines. Looking closely, I see that they are spaced at dash cycle repeat >>length intervals. Instead of that, I think it would be more useful to have a >>line at 25%, 50%, 75% and 100% of the first dash cycle repeat length. In >>other words, each line would mark the beginning of the first complete >>pattern cycle after the origin for one of the lines in the group. They could >>be labeled and/or color-coded, so their purpose is clear. >>Finally, I would like to see a little more explanation in the HTML about >>what exactly is being tested and what to look for in a successful, i.e. >>passing, test. >>HTML Markup Violations >>There are several issues with the HTML markup that must be addressed; e.g. >>there is a CAPTION element as a direct child of the BODY. This element is >>only allowed as a child of TABLE (see >>http://www.w3.org/TR/html401/struct/tables.html#edef-CAPTION). >>WebCGM 2.0 Profile Violations >>There are two profile violations that must be addressed: >>1) The CGM contains a Text Precision element with a value of string (0). >>WebCGM (see T.20.11) only allows a value of stroke (2). >>2) The CGM does not have a Character Set List element in the Metafile >>Descriptor section. WebCGM (see T.16.14) requires this element for all >>metafiles containing graphical text. >> >> >>--------------------------------------------------------------------- >>To unsubscribe from this mail list, you must leave the OASIS TC that >>generates this mail. Follow this link to all your TCs in OASIS at: >>https://www.oasis-open.org/apps/org/workgroup/portal/my_workgroups.php > > >--------------------------------------------------------------------- >To unsubscribe from this mail list, you must leave the OASIS TC that >generates this mail. Follow this link to all your TCs in OASIS at: >https://www.oasis-open.org/apps/org/workgroup/portal/my_workgroups.php > >
[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [List Home]