summaryrefslogtreecommitdiff
Commit message (Collapse)AuthorAgeFilesLines
* configure: properly check for darwin hostdarwin-check-2.50Ignacio Casal Quinteiro2020-12-291-1/+1
|
* build: use -framework Foundation on macOSChristian Hergert2020-12-042-0/+20
| | | | | | | This fixes an issue where link symbols are not available for basic Obj-C message passing and other release/retain operations. Fixes #545
* Post-release version bump to 2.50.3Federico Mena Quintero2020-12-043-35/+34
|
* cargo update2.50.2Federico Mena Quintero2020-11-201-30/+56
|
* Update NEWSFederico Mena Quintero2020-11-051-0/+4
|
* Bump version to 2.50.2Federico Mena Quintero2020-11-053-92/+98
|
* text: Don't create Spans for chars nodes that end up as empty strings after ↵Federico Mena Quintero2020-11-053-13/+34
| | | | | | | xml:space normalization The new test fails without this commit; it creates inter-line spacings in the second <text> element.
* text: Only push a new chunk if the tspan has any of x or y attributesFederico Mena Quintero2020-11-051-8/+7
|
* text: Propagate dx/dy across nested <tspan>Federico Mena Quintero2020-11-051-2/+9
|
* text: normalize some lengths as early as possibleFederico Mena Quintero2020-11-051-48/+46
| | | | | | | Don't keep dx/dy in the structs as Option<Length<...>>; keep them as f64. x/y need to be kept as Option<f64> because they affect chunking.
* Fix the debian:testing_release buildSven Neumann2020-10-051-1/+1
| | | | | | This broke when I changed the build to run on the x86_64 image, but forgot to change the PATH for gdk-pixbuf-query-loaders, see commit c5f09c73b5f76e77fa6d784f54b600996a75c8dc.
* Bump version to 2.50.12.50.1Federico Mena Quintero2020-10-023-3/+3
|
* Update NEWSFederico Mena Quintero2020-10-021-0/+18
|
* Remove references to files that have long been removedSven Neumann2020-10-021-5/+1
|
* Move the render_to_viewport tests to intrinsic_dimensions.rsFederico Mena Quintero2020-10-023-278/+273
| | | | | | All of those are about rendering/measuring with a viewport, and are related to the intrinsic dimensions of the SVG document. So let's have them together.
* Add test with a percentage viewportFederico Mena Quintero2020-10-021-0/+33
|
* Add test with no intrinsic dimensionsFederico Mena Quintero2020-10-021-0/+34
|
* Add test with no width/heightFederico Mena Quintero2020-10-021-0/+31
|
* Add test with a viewBoxFederico Mena Quintero2020-10-021-0/+31
|
* (#554): Do not ignore the viewport in the geometry_for_layer() APIsFederico Mena Quintero2020-10-022-6/+9
| | | | | | Convert the final bounds to be relative to the starting transform. Fixes https://gitlab.gnome.org/GNOME/librsvg/-/issues/554
* Really make the initial viewport have its upper-left corner at (0, 0)Federico Mena Quintero2020-10-021-1/+8
|
* render_layer: make consistent with get_geometry_for_layerFederico Mena Quintero2020-10-021-11/+6
|
* Give a better name to the failing test and adjust itFederico Mena Quintero2020-10-021-2/+2
| | | | We are not ready to test percentage viewports just yet.
* push_coord_units: Always push a Viewport, instead of sometimes returning a ↵Federico Mena Quintero2020-10-021-4/+8
| | | | copy of the last one
* Note that push_new_viewport() also sets a clip rectangleFederico Mena Quintero2020-10-021-1/+1
|
* New helper function to push a full ViewportFederico Mena Quintero2020-10-021-12/+19
|
* Rename view_box_stack to viewport_stack; store full Viewport structsFederico Mena Quintero2020-10-021-17/+21
|
* Add helper DrawingCtx.push_coord_units()Federico Mena Quintero2020-10-023-48/+20
| | | | | | | This takes care of the pattern of pushing a unit-sized viewBox for objectBoundingBox. https://www.w3.org/TR/SVG2/coords.html#ObjectBoundingBoxUnits
* Make another call to push_view_box(1.0, 1.0) consistentFederico Mena Quintero2020-10-021-9/+13
|
* Make a call to push_view_box(1.0, 1.0) consistent with the othersFederico Mena Quintero2020-10-021-1/+4
| | | | We'll replace that little pattern with a helper function.
* Always grow the stack in push_new_viewport()Federico Mena Quintero2020-10-021-3/+3
| | | | | | | We didn't grow the stack if the new vbox would be the same as the last one (i.e. when the passed vbox == None). Now we always grow the stack. The idea is that eventually we'll be able to automatically restore the last viewport's transform on Drop.
* Extract function to get the top viewboxFederico Mena Quintero2020-10-021-3/+8
|
* Define a Viewport structure and use it for the DrawingCtx.initial_viewportFederico Mena Quintero2020-10-021-17/+34
| | | | | | | | For now, this is instead of keeping separate fields for initial_transform and rect (the size of the initial viewport). This indicates that the hack in draw_node_to_surface() is especially pernicious now, since we are monkeypatching the initial_viewport...
* Store a complete ViewBox in ViewParams, not just the sizeFederico Mena Quintero2020-10-022-14/+9
| | | | | We'll move to a model where we pass the current viewport around, which is partly defined by a ViewBox.
* Use a helper function to clip to a rectangleFederico Mena Quintero2020-10-021-22/+10
|
* Don't clone a crFederico Mena Quintero2020-10-021-6/+5
|
* cargo fmtFederico Mena Quintero2020-10-022-20/+9
|
* Make ViewBox opaque; impl Deref<Target = Rect> for itFederico Mena Quintero2020-10-029-32/+50
| | | | That way we avoid `vbox.0.foo` everywhere.
* Docs on ViewBox to remind me that it is a new coordinate spaceFederico Mena Quintero2020-10-021-0/+3
|
* Don't special-case measuring in DrawingCtx::newFederico Mena Quintero2020-10-021-23/+14
| | | | Thanks to Emile Snyder for nothing that that special case is unnecessary.
* WIP: disable that failing testFederico Mena Quintero2020-10-021-0/+1
|
* WIP: Add a failing test for obtaining geometries with an offsetted viewportFederico Mena Quintero2020-10-021-0/+31
| | | | This is for https://gitlab.gnome.org/GNOME/librsvg/-/issues/554
* Move Svg::get_size, a relic of the C API, to handle::get_svg_sizeFederico Mena Quintero2020-10-022-43/+40
| | | | This is only used for the implementation of get_geometry_sub().
* Svg.get_size - Inline get_unnormalized_size into here, so we can extract ↵Federico Mena Quintero2020-10-021-2/+10
| | | | this function
* Remove superfluous conversionFederico Mena Quintero2020-10-021-2/+2
|
* Handle::get_node_geometry_with_viewport - inline in its two callersFederico Mena Quintero2020-10-021-34/+34
| | | | | | The only difference in the two callers was the viewport argument, which is our unit_rectangle() quasi-hack for the measuring case. Let's see if we can refactor that away.
* draw_tree - Take a single DrawingMode enum instead of ↵Federico Mena Quintero2020-10-022-29/+39
| | | | | | | limit_to_stack/node/cascaded arguments Those are in fact related. The cascaded values can be derived from the node at which drawing starts, and the root is just a convenience for the function.
* Create the drawsub_stack outside the DrawingCtxFederico Mena Quintero2020-10-021-7/+12
|
* Create the drawsub_stack in draw_tree(), not in DrawingCtx::newFederico Mena Quintero2020-10-021-11/+7
|
* Combine DrawingCtx::new and draw_from_stack into a single draw_tree functionFederico Mena Quintero2020-10-022-30/+60
| | | | | | | | The DrawingCtx is only used with those two functions from the toplevel handle.rs. So, combine them into a single one that does not return a DrawingCtx at all. This makes DrawingCtx an invisible detail from the viewpoint of handle.rs, and should let us clean up the creation of DrawingCtx more easily.
View Lorry Controller Status