BHFLDCFR6ZOUVAZUOYUSYACWDVY52GAXR6Q4TR4WMCHHFGECXQ3AC
Still reproducible.
K2W4ZPQWK5KCEMR6RWH6AM4K2YEFS6GUZVNP4HXJWSIIZYJTMXRQC
There were two issues indeed, I believe they’re now fixed.
Can you explain why second diff works so long?
$ pijul init p1
$ cd p1
$ cp -Rp /bin/[a-m]* .
$ pijul add -r .
$ pijul record -am"."
Hash: AYQ3ZXDKXWXQ7RUAYKSGVJBMAMN7WXLYAIPSVNDJJ2PFA62JNBIAC
$ time pijul diff -su
real 0m0,296s
user 0m0,079s
sys 0m0,122s
$ touch *
$ time pijul diff -su
real 1m0,151s
user 0m55,230s
sys 0m4,855s
$ time pijul diff -su
real 1m0,423s
user 0m55,592s
sys 0m4,802s
$
RAS4PCNUEODS72Q5LEPVE54AUYWX656ZHTAERRVG2AJ6IHKO4WRAC
Indeed, pijul diff
was read-only (or rather, it didn’t commit the transaction). Now it touches the channel in that specific case, and commits.
Confirmed as fixed.
Real thriller: