My theory is that because p2 no longer produces nor consumes md5,
it will fail recompute these sums if they are already present and
it will fail to notice they are wrong if they are present.
Maybe you are using quite an old version of Tycho with quite an
older version of p2?
I'll investigate completely removing these unused properties
during aggregation; the aggregator already computes stronger
checksums if they are missing.
I changed WTP's contribution to use Oomph's jetty repo and have
rebuilt staging. I see this in the log:
I checked that the staging version of the artifact is not a
signed version, so I think this is fixed.
Regards,
Ed
On 28.08.2023 12:11, Ondrej Dockal via cross-project-issues-dev
wrote:
Hey folks,
in RedDeer build [1] we are facing an issue when running
the tests with a checksum for org.eclipse.jetty.jndi.
Error message:
12:05:07[INFO] Fetching org.eclipse.jetty.util_10.0.15.jar from https://download.eclipse.org/staging/2023-09/plugins/ (557.33kB)
12:05:07[INFO] Fetching org.eclipse.jetty.jndi_10.0.15.jar from https://download.eclipse.org/staging/2023-09/plugins/ (56.39kB)
12:05:07[ERROR] An error occurred while transferring artifact canonical: osgi.bundle,org.eclipse.jetty.jndi,10.0.15 from repository https://download.eclipse.org/staging/2023-09:
12:05:07[ERROR] Problems downloading artifact: osgi.bundle,org.eclipse.jetty.jndi,10.0.15.:
12:05:07[ERROR] MD5 hash is not as expected. Expected: 785f479c6433717bee8e9bb94df56c11 and found 8f2d6b1e2acef3285e3a12f62042890c.