Precision issues with high dpis and drawquad#6
Open
Eolien55 wants to merge 2 commits intoaligrudi:masterfrom
Open
Precision issues with high dpis and drawquad#6Eolien55 wants to merge 2 commits intoaligrudi:masterfrom
Eolien55 wants to merge 2 commits intoaligrudi:masterfrom
Conversation
On some platforms (eg, GCC x86_64), long means 32 bit. This is a problem in pdfpos00, because calculations such as `uh * 1000 * 72` (in pdfunit) are likely to overlow, causing important bugs. This is especially the case when using high dpi values, such as 720000.
The magic formula was wrong, so the circles looked slightly off.
Contributor
Author
|
Hi, rather than replacing long by doubles, I changed it to long long. |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Hello,
This pull request includes 2 commits very useful when using high DPIs (7200 is enough to trigger bugs).
The first makes drawquad more aligned with what a true circle quadrant should like using Bézier curves. Moreover, when using high DPIs, the calculations would overflow (especially the sign test). To fix overflow and precision issues, replacing long's by double's is enough.
The second commit fixes pdfpos and pdfunit. For similar reasons, these 2 functions concentrated much of the overflows of neatpdf. Replacing 32-bit integers by 64-bit might have been enough, but the calculations used to simulate decimal numbers are the source of many rounding errors, which are problematic when precision is needed (eg, when using grap). Replacing the long's by double's is, here too, sufficient.
Thanks for the work you put in your very fine pieces of code,
Regards,
Elie Le Vaillant