Hi @benros, 

Not sure if such parameter is there. Anyhow it's not in the workspace .fmw file as far as I can see. 

However, this is what I think you could do:

1. You can create a Python scripted parameter that returns the name of a logfile to be used in you FME Workspace. Like (in this example I assume a job is not run more than once a day):

from datetime import datetime
return 'mylogfile_'+ datetime.today().strftime('%Y%m%d') + '.log'

2. The workspace should have its log file name linked to the value of that scripted parameter. You can set the log files name from the navigator, under Workspace Parameters > Logging. 

3. From another scripted parameter you can check the available logfiles, and select the last one (before the current run). The other scripted parameter should return the date part from the logfile name found.

4. This last scripted parameter can be used to select your records in the where clause of the database reader. 

No extra transformer needed :-) Not sure if this is neat FME common practise...

from datetime import datetime
return 'mylogfile_'+ datetime.today().strftime('%Y%m%d') + '.log'

Such a parameter might be a nice idea to raise, however.

@benros I think using your own suggestion of a history table or special management table is probably the most robust solution. Adding an extra reader isn't big overhead for FME. The pattern in your workspace would be something like:

Postgresql (history table) -> SQLExecutor or FeatureReader (data tables)

Perhaps you could also ask Snowflake?

https://support.snowflake.net/s/question/0D50Z00008lgahUSAQ/how-to-check-the-last-time-a-table-was-loaded-

I decided to go for a classic solution.

I created an admin table which holds, for each workspace, the maximum value of this increment field, from the last run (first value '2000-01-01 00:00:00').

After writing this data with a Writer, I update this admin table with the new max(value), from the current run, which will be the base value for the next run.

Thanks you all for your answers