Lofton,
thanks for the clarification. If the group thinks that interior-style
is needed I am fine with it.
However, from a script writer’s perspective I wouldn’t know how
to set a pattern or hatch index at all. We are talking about a generic
programmer without CGM knowledge, so it will be difficult at all to know that
interior style allows for specification of patterns and hatchings as well. Then
on top the viewer would expect an index. Other than with Metacheck I wouldn’t
know how to get to such an index (beyond 1..6). For example, in IsoDraw there
is no way to get to this index.
More than that, the various hatch styles and patterns in test
files show a great variety, and I guess that nobody uses these.
When it comes to hatchings, it will be the standard hatch types
45°, -45°, cross-hatch, with variations of line distance, that people would
want to use. Perhaps things would become a lot simpler if we would allow for
these simple hatch types only initially, and document them with pics in the
spec. Patterns is more difficult I guess, no idea here.
Let’s hear what others think, perhaps somebody has a better idea.
Dieter
From: Lofton Henderson
[mailto:lofton@rockynet.com]
Sent: Mittwoch, 14. Januar 2009 15:59
To: WebCGM
Subject: RE: [cgmo-webcgm] How does interior-style work?
Hi Dieter,
I agree with your observation that the present configuration is qualitatively
more complex than 2.0. You are correct that it will be difficult for
script writers to detect programmatically what pattern indices and hatch
indices (other than the default 1..6) are defined in the CGM. A' priori
knowledge is implied. (The ACI allows the hatch styles to be defined, but
not the patterns.)
The interior-style SP has been in the 2.1 drafts from the beginning, and was in
tables that Dave and Stuart put together before the first drafts. Those
tables contained all possibilities, and we whittled down to the ones that
people wanted to keep. I recall past discussion of difficulties such as
you identify, and other similar ones.
Do you (or anyone) want to propose that interior-style be removed?
Majority support would be needed.
All -- if such a proposal is forthcoming, please speak up, pro or con about
interior-style SP (keep or remove)?
Regards,
-Lofton.
At 09:19 AM 1/14/2009 -0500, Weidenbrueck, Dieter wrote:
Lofton,
your description of Interior Style as per CGM:1999 is correct, of course.
However, so far we have tried to keep things simpler in the DOM compared to the
CGM world. We should try to follow that path here as well, and avoid complex
rules for stroked parts.
Also, it will be extremely difficult for a script writer to know hatch or
pattern indices, but there may be no way around this if we allow for different
interior styles.
I was not aware that we did want to specify interior styles as well, I thought
we only wanted to be able to change the color of an already applied style.
Dieter
From: Lofton Henderson [mailto:lofton@rockynet.com]
Sent: Mittwoch, 14. Januar 2009 01:23
To: Bezaire, Benoit; WebCGM
Subject: RE: [cgmo-webcgm] How does interior-style work?
Benoit --
I'll put this on the agenda for Thursday discussion in the WG. (Which
excludes some TC members -- so please reply/comment now on this list if this
topic interests you!) Maybe also next week in the TC.
I'll take a crack at the "general explanation" in the next few days,
unless someone else wants to have at it.
Aside from the question of "side effects", there might be technical
problems with your suggestion to have fill-color, pattern-index, hatch-index
SPs override the CGM's INTERIOR STYLE. For example, FILL COLOUR is used
in more than one way in CGM:1999.
** if INTERIOR STYLE is 'solid', it is the color of the interior;
** if INTERIOR STYLE is 'hollow', it is the color of the border (as distinct
from the edge);
** if INTEROR STYLE is 'hatch', it is the color of the hatch lines;
** it doesn't affect 'geometric pattern', 'pattern', or 'interpolated
interior', IIRC.
So the fill-color SP, at least, would not have an unambiguous association for
overriding INTERIOR STYLE.
I haven't yet thought enough about pattern-index and hatch-index SPs -- whether
there are technical issues with them having the INTERIOR STYLE override side
effect.
About your observation that 2.0 only contained fill-color SP... Good
observation. So fill-color SP would only have an effect if the INTERIOR
STYLE were 'solid', 'hollow', or 'hatch'.
I agree that we should specify test cases and assign them, after we have
endorsed some new text for a general explanation of how SPs work.
Thoughts (anyone)?
-Lofton.
At 09:24 AM 1/12/2009 -0500, Bezaire, Benoit wrote:
A general explanation should suffice. However, I wonder if interior-style
is required at all... should we consider an approach where setting one of the
following SP: fill-color, pattern-index, hatch-index automatically overrides
interior-style? It seems awkward for the script writer to always have to make
two calls, first to set the color or index, the second to change the interior
style.
Please note that in WebCGM 2.0, we added fill-color and there's no mentioning
at all of interior-style?!
Test cases should also cover the general explanation.
From: Lofton Henderson [mailto:lofton@rockynet.com]
Sent: Saturday, January 10, 2009 1:18 PM
To: WebCGM
Subject: RE: [cgmo-webcgm] How does interior-style work?
Benoit,
At 10:41 AM 1/9/2009 -0500, Bezaire, Benoit wrote:
Since you used the words 'opinion' and 'guess'... the working group will have
to work on better defining this.
Apologies, I was rushing and my too-casual wording doesn't reflect that I'm
actually quite certain how it works. As I described:
[[[
order does not matter as long as *both* changes happened before the display
redraw (i.e., you might want to delay updates till you set both). If not,
then given the modal nature of CGM attributes, I would: set pattern
index; change interior style to pattern.
]]]
FIRST order of business: this will be on the Thursday telecon, to endorse
as correct or to refute. If the latter, i.e., if I'm wrong, then we need
specific counter-proposal and rationale from someone.
SECOND. Let me explain why it is correct, and then let's think about how
improve the specification might be better.
(Patience please! ... it has gets lengthy...)
Overview:
----------
1.) Though somewhat intricate, it is nevertheless unambiguous in the CGM:1999
specification how all permutations and combinations of fill attributes would
work in static sequences of metafile elements, in various orders etc.
(Forget about interactive SP-setting for the moment.)
2.) I think it is also unambiguous how setRedraw works to allow or suppress the
immediate effect of individual SP-setting calls.
So if we can devise a general statement/rule that relates how a given
SP-setting call affects the modal attribute list of the APS (see CGM:1999
section 6.7), then CGM:1999 defines the exact graphical effect that should
happen at the redraw after each SP-setting call. Combining that with the
setRedraw state, then such a general statement/rule can be used to answer any
detailed question such as yours.
Detailed Explanation:
---------
I think the specification should describe the relationship between
interior-style and other relevant style properties.
CGM:1999 (see section 6.7) is unambiguous about the graphical effect of static
sequences of elements in metafile instances like:
FILL COLOUR 'red'
INTERIOR STYLE 'solid'
PATTERN INDEX '9'
POLYGON
INTERIOR STYLE 'pattern'
POLYGON
(the first polygon is solid red, the second is pattern #9)
or
INTERIOR STYLE 'solid'
INTERIOR STYLE 'pattern'
POLYGON
PATTERN INDEX '9'
POLYGON
(the first polygon is pattern #1 (the default pattern index), the second is
pattern #9)
If we can understand how each SP-setting call alters the "virtual"
metafile element sequence in an APS, then post-SP-setting effects can be
understood from CGM:1999. WebCGM doesn't need to replicate the details,
for example, of how interior-style and pattern-index DOM SP settings interact,
if we make it clear how each SP setting affects the (virtual) metafile-element
sequence in the target.
The way in which the SP-setting calls affect the metafile element sequence of
the APS shouldn't be hard to describe. Here is a fast, off-the-cuff first
cut:
When the metafile is first interpreted, conceptually a virtual copy of the
elements in the APS is made and is initialized from the actual APS content of
the metafile. Call this the display list for the APS. An SP-setting
function like interior-style (e.g., to 'pattern') has the effect of setting the
INTERIOR STYLE to 'pattern' in the modal attribute list (see CGM:1999 sec 6.7)
, throughout the APS. I.e., regardless of the initial interior style(s)
affecting each graphical primitive throughout the APS, the INTERIOR STYLE is
reset to 'pattern' for all primitives as a result of the SP call. The
interpreter knows how to draw this from CGM:1999, and the interpreter
immediately does redraw it unless there has been a call to
setRedraw('disableAll').
So ... does the order matter? Yes. If you do this to an APS that
has one graphical primitive element -- a polygon as its last element:
setRedraw('enableAll')
setSP(interior-style, 'solid')
*** redraw-1 ***
setSP(interior-style, 'pattern')
*** redraw-2 ***
setSP(pattern-index, 11)
*** redraw-3 ***
Then you get: *** redraw-1 *** solid polygon in the current fill color;
then *** redraw-2 *** pattern polygon with current index (default #1, if it has
not otherwise been set in the metafile); then *** redraw-3 *** pattern polygon
with index #11. (On the other hand, if you disable redraw at first and
re-enable at end, you get only #3.)
setRedraw('enableAll')
setSP(interior-style, 'solid')
*** redraw-1 ***
setSP(pattern-index, 11)
*** redraw-2 ***
setSP(interior-style, 'pattern')
*** redraw-3 ***
Then you get: *** redraw-1 *** solid polygon in the current fill
color; then, *** redraw-2 *** [unchanged] solid polygon in the current
fill color; then, *** redraw-3 *** pattern polygon with index #11.
(Note: the interpreter could detect "no change" and
optimize by not redrawing #2).
Summary:
I claim that all specific questions like yours are answered by the existing
rules of CGM:1999, and need not be answered one-by-one in WebCGM, if this model
relating the aftereffect of a DOM SP-setting call to a sequence of static
metafile elements is clearly understood.
Question:
What should be added to the spec? I vote against situation-by-situation
(element-by-element) explanations, if we can draft a general explanation of the
rules that helps each person to answer such specific questions.
Thoughts?
Regards,
-Lofton.
Benoit.
From: Lofton Henderson [mailto:lofton@rockynet.com]
Sent: Friday, January 09, 2009 10:24 AM
To: Bezaire, Benoit; WebCGM
Subject: Re: [cgmo-webcgm] How does interior-style work?
My opinions...
At 10:13 AM 1/9/2009 -0500, Bezaire, Benoit wrote:
I'm slightly confused with interior-style, more specifically the dependency
that other style properties have on interior-style.
Let's say a WebCGM metafile contains an APS (polygon filled with red color). A
script writer wishes to change the fill to a pattern, how does he do that?
He needs to change the interior-style and the pattern-index, right? Does the
order matter?
My guess: order would not matter as long as *both* changes happened
before the display update (i.e., you might want to delay updates till you set
both). If not, then given the modal nature of CGM attributes, I
would: set pattern index; change interior style to pattern.
Another question is if fill-color only has an impact if interior-style is set
to solid?
Fill color also defines the drawn border when interior style is 'hollow' (as
distinct from the edge, which has its own attributes).
-Lofton.