NOvARawInputSource should not label products as being from "source"
NOvARawInputSource, because of its inheritance from art::RawInputSource, has been made to call produces<T>, and thus to label the products it reads from a file with its "module_label", which is "source".
We should provide a different version of produces<T> (I suggest calling it fakes_production<T>, because the source is not really producing the item, it is just reconstituting it), which accepts an additional argument that is used as the "module_label" with which the product being registered is marked.
Temporary rollback of changes for issue #883 to get NovaRawInputSource working again.
Combined commit: ART-side implementation of issue #883.
Comments from individual commits below.
Add ToyRawInput to test RawInputSource
Improve some error messages
Fix configuration of IntProducer
Clean up layout
Improve format of error message
Improve formatting of error message
Partial implementation and test of reconstitutes
Checkpoing: renaming InputSource
Checkpoint after rebase
Checkpoint: First InputSource as pure interface
Fix typo in installed header name
Move registerProducts out of InputSource interface
Add default implementation for readEvent and skipEvents to InputSource.
Remove doEndRun and doEndSubRun from InputSource.
Remove calls to doEndRun and doEndSubRun
Rename repeat() to repeat_() and make it private.
Remove call to InputSource::repeat()
Fix bad layout.
Complete reworking of public interface.
First skeleton of FileReaderSource
Checkpoint: looking for Brian or Rob
Checkpoint before renaming EventCorruption
Change all references to EventCorruption to DataCorruption instead.
Replace all uses of boost::shared_ptr using do_nothing_policy with exempt_ptr.
"We're not fixing it, we're just using it." Sorry, I had to fix this one:
I have a sensitive stomach.
Not a whole lot of progress, but code compiles and (old) tests pass.
Sort sources for ease of finding things.
Implementation of makeRunPrincipal; two more to go.
Implement remaining functions in PrincipalMaker.
Part 1 of (probably) 2: abstract functionality out of DataViewImpl so it can be called from places that don't inherit from it.
Guard CLHEP use to prevent nullptr troubles.
Handle all the work of putting a product in the event. Lots of stuff still to do here.
Revert "Guard CLHEP use to prevent nullptr troubles."
This reverts commit 09c647b9f5ee1af6519fa41be063a1c4ee457c57.
After testing, we don't have a reliable solution for
avoiding redfining the nullptr macro, so we have to just let it
Checkpoint before splitting ProductPrincipalPutter.
Split ProductPrincipalPutter into two new headers per discussion with MP.
First simple test of ToyRawInput: let the debugging commence!
Improve interaction with ParameterSet and indent.
std::basic_string<char> -> std::string to attempt to trace product problem.
Fix pastos in EventPrincipal and SubRunPrincipal.
Move source signal emission from DRISI and friends into EventPrincipal; retire source construction signals.
Need empty ParameterSet in module description.
Add missing implementation of readFileBlock().
Runs still start at 1 even though the invalid number is still -1.
a bare string product appears to be problematic -- more investigation in a different forum.
Test for bare string as product for further investigation: currently fails.
Fix out-of-date usage message.
Fix behavior of FileReaderSource with respect to moving to the next file.
Round out the tests and put some more sanity checks in FileReaderSource.
Fix erroneous conflict resolution.
Add missing return.
Fix remaining fixable compilation warnings.