[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [List Home]
Subject: RE: re[4]: [cgmo-webcgm] review of aci files
Lofton,
We are closer to agreement than you realize. I also believe (but
unfortunately failed to explictly state in my earlier message), that
metafiles in the test suite should be WebCGM compliant to some specific
profile edition.
Thanks for clarifying the use case for ACI font substitution. It sounds to
me like a single test for the "official" use case is impossible. Since the
font mapping is to "locally available resource" doesn't that make any test
platform dependent? In other words, a test written for the Windows platform
would fail on Mac OS X because the "locally available resource" is
different. Wouldn't you need a separate test for each platform that you
wanted to test on?
Rob
-----Original Message-----
From: Lofton Henderson [mailto:lofton@rockynet.com]
Sent: Wednesday, March 25, 2009 7:08 PM
To: Robert Orosz; 'CGM Open WebCGM TC'
Cc: David Cruikshank (E-mail)
Subject: RE: re[4]: [cgmo-webcgm] review of aci files
Rob,
Thanks for your thoughtful reply. I have some responses.
First, I looked at the Conformance Chapter for guidance:
http://www.w3.org/Graphics/WebCGM/drafts/current-editor-21/WebCGM21-Conf.htm
l#webcgm_conformance_CoP
I don't find any thing definitive. But we are defining conformance for 2.1
things, including 2.1 metafiles, DOM, and ACI. To me, it makes sense that
newly constructed metafiles to test 2.1 features should be 2.1
metafiles. Furthermore, one would need to be really careful about 1.0 or
2.0 metafiles -- if they contained any deprecated or obsoleted features
(e.g., multi-picture, or any of the others), then I believe a conforming
2.1 DOM or ACI would not have to handle them!
Onward to some of your specific points...
At 03:38 PM 3/24/2009 -0600, Robert Orosz wrote:
>Lofton,
>
>Your summary points #1 through #5 below are accurate.
>
>Regarding point #3, I don't recall ever mentioning that new metafiles in
the
>test suite should follow the WebCGM 2.1 profile. In the reviewer
guidelines
>[1], I only mention that the CGMs should be valid without mentioning a
>specific profile edition. This was intentional. My thinking was that the
>CGM need only be valid with respect to the base CGM 99 standard and also be
>suitable for its specific test.
I disagree with this idea. For example, a CGM without a font list is valid
CGM:1999. I don't think we want CGM-valid, but WebCGM-unfriendly,
metafiles in the test suite. I would therefore suggest that the metafile
should be WebCGM compliant to at least *some* version. As I argued above,
take care with 1.0 & 2.0, lest deprecated or obsoleted features are there
(and pass MetaCheck), but 2.1 viewers are not required to handle them!
>For example, some of the "new" tests were
>added to fill gaps in the test suite; i.e. they are testing WebCGM 2.0 or
>even 1.0 features.
I think in these cases -- supplying tests to fill holes in 1.0 or 2.0 parts
of the test suite, I would have no objection to 1.0 or 2.0 WebCGM
ProfileEd's. (But I think it would be good that all such metafile contents
are still required in 2.1 -- i.e., not deprecated or obsoleted.)
>It didn't make a lot of sense to me to require a WebCGM
>2.1 metafile in those cases. However, I did raise the profile issue in one
>of my test reviews [2]. The issue here was not simply the fact that the
>profile was ATA GREXCHANGE instead of WebCGM. The main issue in my mind
was
>that the metafile contained elements that were not allowed in any edition
of
>WebCGM. The re-submitted metafile is a compliant WebCGM 2.0 metafile, and
I
>accepted it without any reservations.
At this point I have a philosophical divergence. We all know that DOM and
ACI will be used to handle non-WebCGM content, or invalid WebCGM
content. However, we had long arguments about this (in the ACI context) in
the past couple of years, and the formal position is this: ACI font sub
(for example) is intended to facilitate matching font names in valid
WebCGM's (either Adobe13, or other-plus-FontProp) to locally available
resource.
It was a spirited debate (Boston, IIRC) that the official use case and
design purpose should not be the handling of invalid WebCGMs, or non-WebCGM
metafiles.
See here,
http://www.w3.org/Graphics/WebCGM/drafts/current-editor-21/WebCGM21-Config.h
tml#ACI-motivation
,
where it says:
"The WebCGM Aplication Configuration Items file (ACI) allows the user to
improve font interchange by specifying a desired font mapping when WebCGM
applications process WebCGM content."
Of course everyone is going to use it that way. And we did throw in the
font-name-normalization bits, despite the formal statement of purpose.
All that said, coming back to the test suite: I think the Test Suite
should honor the intent and good practice. I'm not much of a fan of
throwing files into the 2.1 WebCGM Test Suite which are not valid WebCGM of
*any* version.
>Don's ACI work brings up another consideration that I hadn't initially
>thought about. One of the stated use cases for ACI is to specify things
>such as a line cap style in a version 1 CGM [3]. In a version 3 or 4 CGM,
>this of course is accomplished with the version 3 Line Cap element.
>Presumably this use case is intended for legacy CGMs that predate the
>introduction of version 3 elements.
In the backs of our minds, that was another motivation. But officially, as
stated in your reference [3], we're aimed at: 1.) version 1 WebCGM
instances (still valid in 2.1), altho' I'd be surprised if any such
metafiles exist; 2.) the under-defined 'unspecified' default value of
these things in V3 WebCGM instances (the metafile generator only cared
about any reasonable treatment from amongst the standardized values, but
XYZ Corp wants uniform treatment at presentation time.)
>If so, these CGMs also predate WebCGM
>1.0 by several years. One could then make the argument that a compliant
>WebCGM metafile of any profile edition does not make a good test case for
>these ACI tests.
That's a good point. But it is difficult to reconcile throwing these into
the WebCGM 2.1 test suite, since compliant WebCGM viewers are not required
to handle them.
Idea: a separate demo or samples collection, that show all of the other
wonderful things one can do with these new 2.1 DOM and ACI
features. (Which are the thing a lot of people actually want to do.) And
leave the official 2.1 Test Suite more rigidly aligned the formal stated
purposes and use cases. And consistent with the 2.1 conformance
definitions (Ch.7).
>If you want to look at a specific example, aciFontNormalization.cgm
contains
>Times_Roman as one of the entries in its font list. This is a WebCGM
>profile violation, because Times-Roman (with a hyphen instead of
underscore)
>is specified as a recommended font. However, WebCGM 2.1 specifies rules
for
>"normalizing" font names such as this, and this test is simply intended to
>test that specific subsection of WebCGM 2.1 [4]. The very nature of this
>test makes a WebCGM 2.1 compliant metafile unsuitable.
Right. This is a case where we actually did instanciate one of the real,
behind-the-scenes motivations into the text: "fix trivially deviant but
common usage". I don't see how to avoid testing this feature, without a
metafile that is WebCGM-deviant. (Unless we went back to a plain vanilla
CGM, and I have some discomfort with that.)
>The old CALS CGM profile specified in MIL-D-28003A did in fact specify the
>font name as TIMES_ROMAN with an underscore. Also, a MIL-D-28003A
compliant
>CGM would not contain the WebCGM mandated Font Properties element, because
>that profile did not allow the Font Properties element. In my opinion,
such
>a metafile fits the stated use case and would make a better ACI test case
>than any metafile conforming to any profile edition of WebCGM.
That point is harder for me to argue against. But ... I'd still like to
see most files in the test suite at least *pretending* to be
WebCGM-compliant. For example, in the transparent-cell-colour test,
changing the profile id to WebCGM (from ATA).
It is possible in the font-substitution area (specifically, the
name-normalization rules), that we might deviate from that general
principle. (Still thinking about that one.)
Regards,
-Lofton.
>[1] ftp://ftp.cgmlarson.com/ReviewerInstructions.html
>[2] http://lists.oasis-open.org/archives/cgmo-webcgm/200901/msg00003.html
>[3]
>http://www.w3.org/Graphics/WebCGM/drafts/current-editor-21/WebCGM21-Config.
h
>tml#ACI-motivation
>[4]
>http://www.w3.org/Graphics/WebCGM/drafts/current-editor-21/WebCGM21-Config.
h
>tml#ACI-maplist
>
>-----Original Message-----
>From: Lofton Henderson [mailto:lofton@rockynet.com]
>Sent: Saturday, March 21, 2009 1:56 PM
>To: 'CGM Open WebCGM TC'
>Subject: RE: re[4]: [cgmo-webcgm] review of aci files
>
>
>Don -- the action item is back to you to correct the metafile errors from
>Rob and Ulrich. Can you fix these problems? (Matrix [1] is updated.)
>
>Rob, Ulrich, All --
>
>Does this accurately summarize the metafile problems, as detected by
>MetaCheck (and reviewers)?
>
>1.) Max VDC Extent violation;
>2.) "ColourClass:colour" problem (MetaCheck bug);
>3.) ProfileEd is 2.0 (ought to be 2.1);
>4.) Max Colour Index violation;
>5.) Font List violation ('arial' in Font List but no Font Properties
>element).
>
>Overall opinion: While these problems must be fixed, they are in some
>sense "formal", and I think most of them will not affect an implementor's
>ability to use the file. The overall structure and content of the test is
>appropriate. Details:
>
>#1: the MaxVdcExt element is a pain in the butt! Should never have been
>added to CGM:1999. But ... there it is, so may as well bite the bullet and
>make it right. (IMHO, this is purely a "formal" error -- should have no
>practical impact on early users of the test.)
>#2: metacheck bug, nothing need be changed.
>#3: ought to be 2.1 for new metafiles in the 2.1 test suite, yes? (See
>Rob's comments below about this -- MetaCheck will gripe but we know it's
>not a problem.)
>#4: Another "formal" violation -- I would expect minimal impact on early
>users of the test.
>#5: This is somewhat at the heart of the test. While it *might* not
>impact early users of the test, on the other hand it might. It could be
>fixed by either changing the font to 'helvetica', or by adding a Font
>Properties element.
>
>Thoughts (anyone)?
>
>Btw, once these metafile problems are resolved, we will have a big batch of
>tests ready for "Approved", I think.
>
>Regards,
>-Lofton.
>
>[1] ftp://ftp.cgmlarson.com/test-matrix.htm
>
>At 05:02 PM 3/19/2009 -0600, Robert Orosz wrote:
> >Don,
> >
> >The binary CGMs that you've submitted are invalid. I've attached a ZIP
file
> >containing the MetaCheck trace output for each.
> >
> >Lofton,
> >
> >Please reset the status of these tests to reworking.
> >
> > aciDefaultFont
> > aciEdgeAttr
> > aciFontNormalization
> > aciGenericFonts
> > aciHatchStyleDef
> > aciLineAttr
> > aciLineEdgeTypeDef
> > aciMapListOrder
> > aciMitreLimit
> > aciRestrictedTextType
> > aciSimpleFontSubsitution
> >
> >The CGM conformance violation is the same in each; namely, the VDC Extent
> >exceeds the specified Maximum VDC Extent. Each CGM has (1,1) (-1,-1) for
> >the Maximum VDC Extent and that is well exceeded by the VDC Extent in
each
> >case.
> >
> >These CGMs also highlight an inconsistency in the WebCGM profile. T.16.8
> >allows for 16-bit color index precision. However, T.16.9 caps the
maximum
> >color index at 255, i.e. effectively limiting you to only 8-bit color
index
> >precision. We should make these two table rows in the PPF consistent
with
> >each other.
> >
> >More later,
> >
> >Rob
> >
> >-----Original Message-----
> >From: 'Don Larson (E-mail)' [mailto:dlarson@cgmlarson.com]
> >Sent: Thursday, March 19, 2009 7:37 AM
> >To: Lofton Henderson
> >Cc: 'CGM Open WebCGM TC'
> >Subject: re[4]: [cgmo-webcgm] review of aci files
> >
> >
> >Lofton,
> >
> >I have corrected and updated the following ACI files:
> >
> > aciFontNormalization.aci
> > aciLineAttr.aci
> > aciHatchStyleDef.aci
> > aciMitreLimit.aci
> >
> >
> >Also all the ACI related CGMs have been converted to binary:
> >
> > aciDefaultFont.cgm
> > aciEdgeAttr.cgm
> > aciFontNormalization.cgm
> > aciGenericFonts.cgm
> > aciHatchStyleDef.cgm
> > aciLineAttr.cgm
> > aciLineEdgeTypeDef.cgm
> > aciMapListOrder.cgm
> > aciMitreLimit.cgm
> > aciRestrictedTextType.cgm
> > aciSimpleFontSubsitution.cgm
> >
> >
> >Regards,
> >Don
> >
> > > At 10:22 AM 3/17/2009 -0500, Forrest Carpenter wrote:
> > > >Don,
> > > >
> > > >aciMitreLimit aci file <mitreLimit limitVal=5 />
> > > > should be <mitreLimit limitVal="5" />
> > > >CGM file is clear text, should be binary
> > > >
> > > >aciRestrictedTextType aci file OK
> > > >CGM file is clear text, should be binary
> > > >
> > > >aciLineEdgeType aci file OK
> > > >CGM file is clear text, should be binary
> > > >
> > > >aciHatchStyleDef aci file </hatchStyleDef>
> > > > should be />
> > > >CGM file is clear text, should be binary
> > > >
> > > >aciLineAtt aci file <lineCap lineCapInd=2 lineDashInd=3 />
> > > > <lineJoin lineJoinInd=4 />
> > > > <lineTypeCont lineContMode=3 />
> > > > Should be <lineCap lineCapInd="2" lineDashInd="3" />
> > > > <lineJoin lineJoinInd="4" />
> > > > <lineTypeCont lineContMode="3" />
> > > >CGM file is clear text, should be binary
> > > >
> > > >aciEdgeAtt aci file OK
> > > >CGM file is clear text, should be binary
> > > >
> > > >aciFontNormalization aci file <fontmap> and </fontmap>
> > > > should be <fontMap> and </fontMap>
> > > >CGM file is clear text, should be binary
> > > >
> > > >Regards,
> > > >Forrest
> > > >
> > > >
> > > >-----Original Message-----
> > > >From: Don Larson (E-mail) [mailto:dlarson@cgmlarson.com]
> > > >Sent: Tuesday, March 17, 2009 9:02 AM
> > > >To: Lofton Henderson
> > > >Cc: CGM Open WebCGM TC
> > > >Subject: re[2]: [cgmo-webcgm] review of aci files
> > > >
> > > >Lofton,
> > > >
> > > >All of the following ACI files have been updated by adding-
> > > > <?xml version = "1.0" ?>
> > > > <!DOCTYPE webcgmConfig SYSTEM "webConfig.dtd">
> > > >
> > > >
> > > >aciFontNormalization
> > > >aciGenericFonts
> > > >aciMapListOrder
> > > >aciLineAttr
> > > >aciEdgeAttr
> > > >aciLineEdgeTypeDef
> > > >aciHatchStyleDef
> > > >aciEdgeAttr.aci
> > > >aciHatchStyleDef.aci
> > > >aciLineAttr.aci
> > > >aciLineEdgeTypeDef.aci
> > > >aciMitreLimit.aci
> > > >aciRestrictedTextType.aci
> > > >
> > > >Regards,
> > > >Don L.
> > > >Larson Software Technology
> > > >www.cgmlarson.com
> > > >
> > > >
> > > > > Don,
> > > >
> > > > > When you have corrected this, on every ACI file, could you
>please
> > > inform
> > > > > the TC via the TC list?
> > > >
> > > > > Then I'll update the matrix and we'll be ready for final steps
> >toward
> > > TC
> > > > > "Approved".
> > > >
> > > > > Thanks,
> > > > > -Lofton.
> > > >
> > > > > At 02:57 PM 3/16/2009 -0500, Forrest Carpenter wrote:
> > > >
> > > > > Lofton,
> > > > >
> > > > > None of the aci files in the test suite are valid xml files.
Don
> >has
> > > > > corrected one error I pointed out in his latest update but all
>aci
> > > files
> > > > > in the test suite are missing the two lines below, the first
>line
> >is
> > > > > required and the second line should be included
> > > > >
> > > > > <?xml version = "1.0" ?>
> > > > > <!DOCTYPE webcgmConfig SYSTEM "webConfig.dtd">
> > > > >
> > > > > Regards,
> > > > > Forrest
> > > > >
> > > > >
> > > > >
> > > > >
> > > > >
> > > > > From: Lofton Henderson [mailto:lofton@rockynet.com]
> > > > > Sent: Monday, March 16, 2009 1:33 PM
> > > > > To: Forrest Carpenter; 'WebCGM'
> > > > > Subject: Re: [cgmo-webcgm] review of aci files
> > > > >
> > > > >
> > > > >
> > > > > Forrest,
> > > >
> > > > > At 09:41 AM 3/10/2009 -0500, Forrest Carpenter wrote:
> > > >
> > > > >
> > > > > I don t believe the aci files provided in the tests are valid.
I
> >have
> > > > > attached an example of what I believe the files should look
>like.
> >I
> > > have
> > > > > also found one error in the DTD, dashlength should be changed
to
> > > > > dashLength.
> > > > >
> > > > >
> > > > > Can you be specific, what files you are referring to? I.e.,
>list
> > > them?
> > > > > You had the Action to review these 7:
> > > >
> > > > > aciFontNormalization
> > > > > aciGenericFonts
> > > > > aciMapListOrder
> > > > > aciLineAttr
> > > > > aciEdgeAttr
> > > > > aciLineEdgeTypeDef
> > > > > aciHatchStyleDef
> > > >
> > > > > Does your comment apply to all 7 of these?
> > > >
> > > > > Then Don sent an update message saying he had updated this
> >different
> > > set
> > > > > of files:
> > > >
> > > > > aciEdgeAttr.aci
> > > > > aciHatchStyleDef.aci
> > > > > aciLineAttr.aci
> > > > > aciLineEdgeTypeDef.aci
> > > > > aciMitreLimit.aci
> > > > > aciRestrictedTextType.aci
> > > >
> > > > > I'm getting a little lost, trying to track the status of
>"Review"
> >or
> > > > > "Reworking", and who has the action item, in the matrix.
> > > >
> > > > > -Lofton.
> > > >
> > >
>---------------------------------------------------------------------
>
>
>---------------------------------------------------------------------
>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]