Fix integer overflow in leap year calculation #1955
Conversation
kevinbackhouse
force-pushed
the
FixIssue1954
branch
from
c6ad0d9
to
112d24d
Compare
Codecov Report
@@ Coverage Diff @@
## main #1955 +/- ##
=======================================
Coverage 61.13% 61.13%
=======================================
Files 96 96
Lines 19068 19054 -14
Branches 9745 9732 -13
=======================================
- Hits 11657 11649 -8
+ Misses 5093 5092 -1
+ Partials 2318 2313 -5
Continue to review full report at Codecov.
|
|
Indeed, see https://en.wikipedia.org/wiki/Leap_year#Gregorian_calendar (and the "Algorithm" subsection). Maybe the comment should also say something like "Assuming Gregorian calendar since 1582; years before are assumed to be using the proleptic Gregorian calendar and years <1 are assumed to be using astronomical year numbering" |
|
See also https://en.wikipedia.org/wiki/ISO_8601#Years basically saying the convention for any year before 1582 has to be clearly agreed/identified between communicating parties. |
|
If we get a contributor from Vatican City, we could ask her to deal with this. |
…l?id=39555 Remove bogus code.
kevinbackhouse
force-pushed
the
FixIssue1954
branch
from
112d24d
to
4bc4757
Compare
Fixes: #1954
Credit to OSS-Fuzz: https://bugs.chromium.org/p/oss-fuzz/issues/detail?id=39555
This code looks completely bogus to me. I can't find any reference online that says that negative years should be treated different in a leap year calculation. So the simplest solution to the integer overflow bug is to comment this code out.