Closed Bug 296763 Opened 20 years ago Closed 18 years ago

2d rendering context: beginPath() should call moveTo(0,0)

Categories

(Core :: Graphics: Canvas2D, defect)

Product:
Core
Component:
Graphics: Canvas2D
Type:
defect
Priority:
Not set
Severity:
normal

Tracking

()

Status:
RESOLVED FIXED

People

(Reporter: ted, Unassigned)

References

(
URL
)

Details

The WHAT-WG spec says beginPath() calls moveTo(0,0).
Hrrrm.  I'll have to check what Safari does; this sounds like an error in the spec.

So beginPath() would continue from where the last one left off? Would that be
part of the save/restore state? I think this assertion came from the Apple docs
originally, let me know if it needs to change.
Wait, are the specs being molded to suite what Safari already has?  I always
thought the specs were supposed to be the ones followed, and if another browser
disobeys them, we shouldn't emulate them.
The specification is partly based on Safari's implementation as they implemented
it before there was a specification for the element.
Which begs the question why we are implementing an unfinished spec that has
obvious conflicts with the implementation it's based on.  I remember several
cases in the past where mozilla decided not to implement a spec because it was
incomplete or wasn't a standard yet.  Changing the rules mid-spec also makes a
mess for other applications wishing to be ahead of the game and implement early.
 I know, that's out of the scope of this bug, but it's issues that should be
addressed.
It's like this:
 * Safari invented and shipped <canvas>.
 * Firefox 1.1 will be shipping with a compatible <canvas>.
 * Opera will be shipping a compatible <canvas> in due course too.

The spec is to allow vendors to agree on who is "correct" once two of them have
shipped, and to allow other vendors to implement the same thing without having
to reverse engineer other browsers. While there is only one implementation (as
with <canvas> at the moment), that implementation is king, since we all want to
be compatible. There's no point having a spec that doesn't match reality.

Similarly, the parts of the spec that only WinIE6 implements will be made
compatible with WinIE6.
*** Bug 323999 has been marked as a duplicate of this bug. ***
Fixed by 333177.
Status: NEW → RESOLVED
Closed: 19 years ago
Resolution: --- → FIXED
Nice but it has broken many people's work out there. Did it worth it? I would understand adding new features but breaking it to emulate safari to 100% doesnt feel right. Instead of breaking it, advising people to call moveTo(0,0) after beginPath() would have been more appropriate.

Drawing a perfect circle with a linewidth more than 2 is almost impossible with this change. I cant interpolate the moveTo location, there is always a deformation somewhere. Is there an example to draw a perfect circle with linewidth 3 and radius 5 using this new change? It works with linewidth 1 there is no doubt about it.

If you ask me if the path begins with an arc it shouldnt moveTo(0,0), and who knows if safari doesnt do just that under the hood ;).
You know, based on what Safari does, this is bogus; there should be no implied moveTo(0,0).  I thought I had tested this in Safari first, but I guess not.
As this -does- break web content, I'm going to revert this change.
Status: RESOLVED → REOPENED
Resolution: FIXED → ---
Er, opening  bug 346005 for the revert instead.
Status: REOPENED → RESOLVED
Closed: 19 years ago → 18 years ago
Resolution: --- → FIXED
FWIW, Ian has changed the spec. to follow Safari's implementation so there's no longer an implicit moveTo(0,0).
*** Bug 323222 has been marked as a duplicate of this bug. ***
You need to log in before you can comment on or make changes to this bug.