Athens Re: TxText model
Locked
 
|
2016-04-06 16:58 GMT+02:00 Igor Stasenko <[hidden email]>:
As you are here and in a good mood for feedback, can you please answer some Athens questions or review some code changes / issue entries? I added some instance variables for AthensCairoPatternPaint (origin, extent) This was needed for converting Morphics GradientFillStyle to Athens pattern paint. But I am unsure if this was the intended way to use them. I am interested on issue 14800 AthensCanvas clipRect oddity (Aliaskei thinks the current behavior is on purpose ,but I am not sure). AthensPathSegment: I understand how they use the visit: messages but I don't understand the #accept: or the #sendCommandTo: methods, are they used? Aliaksei fixed a bug in CairoFreetypeFontRenderer>>#glyphsOf:from:to: (the bug was: sometimes, some characters uses glyphs with wrong size). I think some AthensText classes are obsolete: AthensTextScanner/AthensParagraph/AthensTextMorph/AthensPluggableTextMorph Can we remove them?
|
|
|
On 7 April 2016 at 10:32, Nicolai Hess <[hidden email]> wrote:
There are protocols for creating gradient paints in Athens: createLinearGradient: aColorRamp start: aStartPoint stop: aStopPoint createRadialGradient: colorRamp center: aCenter radius: aRadius focalPoint: fp how you storing things inside or transform this information is orthogonal. From that perspective, anything what happens in AthensCairoPatternPaint is implementation detail. If you think that this protocol is not sufficient to define gradients, then fix it.. What happens on concrete implementation is last thing that should worry about. From my POV , two points and color ramp is enough to define linear gradient. Things like extent is a derived data, that can be obtained via simple calculations from original data. That why it wasn't there in a first place.
The intent was to make clipping work with Morphic.. This is true that morphic needs cliprect to be smart and not try to draw anything if it is fully outside clipped areas.. So, this is purely for optimization purpose. But i would prefer to have a consistent clipping model first, i.e.: - clipping using local coordinates - you can use any shape for clipping and then think how to adapt users with their concrete cases. That may, indeed make accessing such properties like clipRect impossible to get, since clipped area cannot be represented by a simple rectangle anymore. But then you could, perhaps instead provide a protocol to tell if given point/bounds/shape is fully outside currently clipped region(s) or not. And thus, you don't have to worry about cliprect anymore.
The intent is to use it for path conversion/transformation etc. It is because we cannot guarantee that all backends will support various kinds of path segments, that Athens providing to user. And so, we need a small private toolset for converting some segments into simpler form, in case of need. So , it is there when you need to iterate over path for things like converting it to polygon etc. But that part needs review and cleanup, because it smells...
Well, if i remember, there are some leftovers from attempt to make text rendering it at least partially work with Balloon backend. While some other is result of failed attempt to make anything with TextMorph. So, nobody will cry if you clean this mess. Best regards,
Igor Stasenko. |
|
|
On Thu, Apr 7, 2016 at 11:15 AM, Igor Stasenko <[hidden email]> wrote:
In Sparta I use cairo_clip_extents to get current clip rect in local coordinates. Together with cairo_path_extents, implementation of AthensCanvas#>>#isVisible: becomes trivial even for arbitrary path (+ assuming that Rectangle is polymorphic with Path). Btw, for simple paths (circle, ellipse, text, rectangle) it is an overkill to ask cairo for path extent that is why it makes sense for optimisation to have subclasses of Path like EllipsePath, PolygonPath and calculate extent in Pharo. If user uses Factory than extent detection degrades to cairo_path_extents. So, in Sparta (now it is still BlPath) there is Path>>#bounds that by default uses cairo_path_extents, still allowing user to override it and implement optimised Pharo version :) Cheers, Alex |
|
|
In reply to this post by Igor Stasenko
2016-04-07 11:15 GMT+02:00 Igor Stasenko <[hidden email]>:
Hm, I still don't know if the outcome was intended. If you compare the two example codes, the first defines a cliprect translates the path transform draws a rectangle (the green one), this rectangle is clipped (that is, the cliprect is not translated - OK) The second example does the same, but defines again a cliprect after the path transform. Now the rectangle is not clipped (by the first clip rect), although the comment in AthensCanvas says "Note that the clipping rectangle is intersected with the currently active clipping rectangle" |
|
|
In reply to this post by Aliaksei Syrel
2016-04-07 11:52 GMT+02:00 Aliaksei Syrel <[hidden email]>:
Do we need an AthensCanvas>>#isVisible: ?
Even if it would not speed up extent calculations, I would prefer too, to have EllipsePath, PolygonPath, ..., It is much more convenient to work with explicit classes than to define everything with a Path, even if everything is just a Path :-)
|
|
|
On Thu, Apr 7, 2016 at 12:16 PM, Nicolai Hess <[hidden email]> wrote: Do we need an AthensCanvas>>#isVisible: ? Yes. There is no need to draw morphs that are outside of current clip area. That is why we have: fullDrawOnSpartaCanvas: aCanvas . . . Even if it would not speed up extent calculations, I would prefer too, to have EllipsePath, PolygonPath, ..., It is much more convenient to work with explicit classes than to define everything with a Path, even I too :) It also helps during debugging and inspection, having that we can mold tools with custom presentations for Path. Cheers, Alex |
|
|
In reply to this post by Nicolai Hess-3-2
On 7 April 2016 at 13:10, Nicolai Hess <[hidden email]> wrote:
Ideally that should work as: - you clip in local coordinates - then you can translate/transform coordinates again - and again clip, but in new coordinates, but that still take into account the outer clipping. IIRC, Cairo doesn't allows nesting for clipping.. but i may be mistaken. Best regards,
Igor Stasenko. |
«
Return to Pharo Smalltalk Developers
|
1 view|%1 views
| Free forum by Nabble | Edit this page |