Huw Davies <[email protected]> wrote: > Perhaps the original commit was added to work around this issue?
No, it was added for a not related reason. > Note, you'll probably have to retrieve the metrics at the design size (2048) > in order to get accurate values for these members. Simply up-scaling the > values > obtained at 100 ppem will lead to rounding errors. As you say, tests are > needed. Yeah, the tests are needed. For now I just wanted to make gdiplus not rely on broken behaviour. -- Dmitry.
