[cairo] CAIRO_STATUS_NO_MEMORY when passing incorrect scaling

[Behdad Esfahbod]

Peter Clifton wrote:
> Hi guys,
> 
> I noticed, as I was debugging some drawing code, that after a certain
> error in my code, all further cairo output was stopped.
> 
> I tracked the problem down to an incorrect scale factor I was passing to
> cairo_scale(), I was in fact passing the X scale factor as 0. due to a
> corner case being hit at a certain zoom factor.

Yeah, that's a bug that we need to fix.  Scaling down to zero should not be an
error.

> What puzzled me in trying to track this down, is that the surface's
> error state was CAIRO_STATUS_NO_MEMORY (at least,
> cairo_status_to_string() reported no memory).

That sounds wrong.  You shouldn't get an error on the surface at all.
Should get it on the context.

> Is this the expected error status for this kind of problem with the
> surface transformation matrix? Should this error be recoverable?

Cairo errors are for programmer and irrecoverable errors.  Degenerate matrices
simply should not cause any error.


behdad

> In my case, the cairo context became basically useless after that one
> mistaken operation, in spite the fact that I reset the CTM afterwards
> with a cairo_identity_matrix() call, all further drawing operations
> failed.
> 
> I've since fixed my code, but wondered whether the behaviour I saw
> constituted a bug which needed reporting.
> 
> Best regards,
>