src|cmake: Fix some FromBytes decoding issues, add more test cases
#253
Conversation
xdustinface
force-pushed
the
pr-fix-from-bytes
branch
from
e9f60d1 to
eebf526
Compare
Test all prefix values, not only 0-10.
xdustinface
force-pushed
the
pr-fix-from-bytes
branch
from
eebf526 to
896e33c
Compare
xdustinface
changed the title
FromBytes for non-infinity zero only buffersFromBytes decoding issues, add more test cases
xdustinface
changed the title
FromBytes decoding issues, add more test casesFromBytes decoding issues, add more test cases
|
Okay, turned out that there are more "first byte values" where it should pass because the 5 lower bits of the first byte are also used as data. I adjusted the test and fixed an issue in the zero bytes check.. now it should have the same behavior like before the relic bug-fix which introduced the change of the behavior. But at the end i guess this should also be fixed in relic itself @dfaranha? Another thing i realized is that with this commit relic-toolkit/relic@1620a03 we get more invalid cases, see 1fa53dc which is required after that commit . I miss some background here to judge if all of the values in here with zeroes appended should be decoded to valid G1 elements. Can either of you @mariano54 or @dfaranha share your thoughts? Also just for the record, this PR now bumps relic to relic-toolkit/relic@b7b2266. If someone has any objections let me know, i can then drop 1fa53dc and 896e33c. |
…0acd68ed9c88d7cd82ec5490ce4
Which is currently the latest `main`
xdustinface
force-pushed
the
pr-fix-from-bytes
branch
from
896e33c to
916e4de
Compare
|
Currently on holidays, but will catch-up tonight! |
|
I have not reviewed this yet |
|
ah i saw gene's review and thought it had gone through |
|
Sorry - however it did pass Guido's testing and it looked otherwise correct - which is why I approved. |
…hia-Network#253) * test: Expand invalid G1/G2 element test cases Test all prefix values, not only 0-10. * util: Introduce `HasOnlyZeros` to check a `Bytes` object for zero bytes * elements: Don't accept zero only buffer with valid prefix in `FromBytes` * test: Drop cases which become invalid with relic commit 1620a03b388e50acd68ed9c88d7cd82ec5490ce4 * cmake: Pin relic to b7b2266a0e4ee6f628f61d3ab638f524a18b52f1 Which is currently the latest `main`
This is a (i think the proper) fix for #251 which basically has been introduced by a bug-fix in relic, see relic-toolkit/relic@c7177c8 and relic-toolkit/relic#206. Please also have a look here @dfaranha :)
a90d56a expands the test case to cover the issue detected in #251 and e9f60d1 is the actual fix.