changing histogram setup causes Whizard to discard an existing event sample

[fbach]

Changing either the setup off an existing histogram or adding a new histogram unrelated to the actual analysis both makes Whizard discard existing events.

[fbach]

Also a change in the analysis makes Whizard discard events. Try

analysis = record pt (eval Pt [H])

instead of filling rap in the example.

[kilian]

I don't get it. What's wrong?

Could you attach the .sin file which doesn't behave as expected?

[Juergen Reuter]

There is an attachment ...

[kilian]

Replying to jr_reuter:

There is an attachment ...

Yes, and it behaves as expected, doesn't it? What do I have to modify, exactly?

[Juergen Reuter]

Change the two lines (one being active in the first run, the second being active in the second run):

histogram rap (-5, 5, 0.5)

and

histogram rap (-5, 5, 0.2)

Then you find in the second run:

| Events: reading from raw file 'proc.evx'
| Events: parameter mismatch, discarding old event set
| Events: closing raw file 'proc.evx'
| Events: generating new events

But it shouldn't regenerate ...

[kilian]

It looks like the histogram is not responsible at all ... I get a parameter mismatch for a modified test even if the second .sin file is identical.

Something's random in the MD5 sum source, probably.

[kilian]

Arghh! The MD5 sum recorded by VAMP includes the output from the integration timer!

[kilian]

Presumably fixed in r5575. Maybe someone can double-check ... it works for me now.

[Juergen Reuter]

I rechecked this. Seems to work. If we find a counterexample, we could reopen. Closing for now.