Wrong assertion in suitesparse.cc about ceres-solver HOT 4 CLOSED

Comment #1 originally posted by sandwichmaker on 2012-06-13T16:02:35.000Z:

You are right this is an off by one error. I will look at this later today.
Thank you for reporting, and I am surprised that this escaped my testing. I will add a test to the effect too.

from ceres-solver.

Comment #2 originally posted by sandwichmaker on 2012-06-14T19:54:06.000Z:

I saw that commit 469bf39 tries to fix this. However, the DCHECK was only part of the problem. The code now happily dereferences row_block_starts.end() in the following if-statement, which results in undefined behavior. I failed to mention that this has always been the case in Release mode. It is necessary to explicitly check for end().

from ceres-solver.

Comment #3 originally posted by sandwichmaker on 2012-06-14T20:11:13.000Z:

ugh, my bad, this is what happens when I do things late night.
I will fix this in a follow up. Thanks for your careful look at his.
Sameer

from ceres-solver.

Comment #4 originally posted by sandwichmaker on 2012-06-15T22:04:20.000Z:

Hopefully fixed and pushed to master for real this time :).

from ceres-solver.