+1 agreed
As for the ignored properties warning, one option without reverting the code
would be to add "ignoreDFDLProperty" to the default list of suppressed warnings.
Though if anyone sets the suppressSchemaDefinitionWarnings tunable, they'll
probably see the excessive and incorrect warnings.
Alternatively, we could just comment out the one or two lines that actually
create the warning, so everything else stays the same, including doing the work
to try to detect warnings, we just don't output anything.
Though, if we can't fix this we may want to just consider reverting the commit
until there is more time to fix it. The old behavior wasn't perfect, but it did
an okay job.
Also, for reference, here's the list of PR's that we should probably work to get
into this release, all of which I think just need another +1 or have relatively
minor changes needed:
#1391 Convert BigInt/Decimal to expected types for packed numbers
#1389 Add new daffodil-tdml-junit library
#1337 Parse Error on Out of Range Binary Integers
#1321 Warn about Multiple Child Elements with Same Name
Dependency updates:
#1398 Update sbt, scripted-plugin to 1.10.7
#1394 Bump actions/setup-java from 4.5.0 to 4.6.0
#1393 Bump codecov/codecov-action from 4.6.0 to 5.1.2
#1390 Update jline to 3.28.0
#1388 Bump msys2/setup-msys2 from 2.24.1 to 2.26.0
#1387 Update scallop to 5.2.0
#1385 Bump SonarSource/sonarcloud-github-action from 3.1.0 to 4.0.0
#1379 Update commons-io to 2.18.0
Have I left anything out or anything else someone was hoping to get in this
release?
- Steve
On 2025-01-03 11:02 AM, Mike Beckerle wrote:
+1 for a release "real soon now".
As for what we need to fix before that release, there are a number of
outstanding tickets related to the new warning mechanism about ignored
properties.
We really shouldn't roll forward with these unresolved issues.
Is there a switch or small change we can make to disable this
mechanism (or have it off by default) without rolling back the actual
commit(s)?
I do expect these ignored property warnings will help users isolate
schema bugs, some of which are undetected in current schemas. It's an
important feature for us to have. But the open tickets indicate it
isn't correct nor complete yet.
There is so much other good stuff on the main branch that I think it
is well worth creating 3.10.0 without the ignored property warnings
feature.
See:
https://issues.apache.org/jira/browse/DAFFODIL-2875 - shows that
shared groups and type definitions in the schema compiler defeat the
current algorithm for detecting unused properties.
https://issues.apache.org/jira/browse/DAFFODIL-2964 - we get too many
warnings from the regression suite - likely indicates we're getting
false warnings? (could be same issue as DAFFODIL-2875, or something
else.)
https://issues.apache.org/jira/browse/DAFFODIL-2798 - ignored property
detector doesn't detect enumeration facet unused properties.
On Fri, Jan 3, 2025 at 10:26 AM Adams, Joshua
<jad...@owlcyberdefense.com> wrote:
All,
It's been roughly 3 months since the release of Daffodil 3.9.0 and a number of
bug fixes and some performance improvements have been added in the mean time,
so I would like to discuss beginning the release process for Daffodil 3.10.0.
There are a few outstanding PR's that we should be able to get merged next week
and then we can hopefully begin the release process on the following Monday
(1/13/25) and allow for a release on the 16th if things go smoothly. Are there
any other significant issues or improvements that we want to get included in a
3.10.0 release?
I'm volunteering to be the release manager this go around.
Josh Adams
