Its about time .. a better date parameter handling in PRD-3.7

For a long time date-parameters were probably the most creepy new feature that we had since finishing the Citrus release a year ago. Unlike a text or a number, the absolute value of dates is dependent on the runtime environment.

Yes, you may have guessed it: Timezones – I just hate them. Not that they give you jetlag when traveling, they also mess up date parsing when you are not very very careful.

First, Java does not have a sane date system. It basically just inherits the low-tech mess of C/C++ and tries to disguise that by calling the milliseconds since 1.1.1970 a “Date” object. But at the end of the day you don’t deal with dates (the day-month-year part, not the time within a day) – you deal with timestamps.

And thus the date “2010-05-30” parsed with the UTC timezone yields something different than parsing the date “2010-05-30” in “Eastern Time”. And if you’re not careful your database will happily jump on the wagon and will return a few hours more or less of your data. I can tell you: the IRS loves randomized balance sheets.

Up until now, the parameter handling for date parameters was as random as it could be. In theory, date parameters are supposed to be passed in as standardized ISO formats. The ISO format yyyy-MM-dd'T'HH:mm:ss,SSSZ is used to transport timestamps in a locale and timezone independent fashion. On the BI-Server, however, dates were always passed around using the format ‘yyyy-MM-dd’ ignoring all time or timezone information. This made it impossible to use time or timestamp parameters there. Ultimately this led to case PRD-2624. Cutting off user input is bad – but it is worse when such bugs stay for a while, as at that point people develop workarounds which (due to the magic of “stay backward compatible and do not break existing reports”) block us from implementing a clean fix. So we create a workaround for the workarounds. Bah! Complexity is the enemy of a good design.

Beginning with PRD-3.7 and BI-Server 3.7 date parameters for PRPTs follow a better schema. Each date-parameter can define whether the date given is a client-side, a server-side date or whether it should be interpreted in a fixed timezone (UTC, for instance) regardless of the client’s or server’s location.

(XActions and all other components work around that problem by declaring everything as string and thus offload the problem into the user space. Use JavaScript to parse your date. The target audience there are sys-admins and technically skilled users anyway – they are used to pain.)

Existing reports now parse the time-stamps correctly and thus at least enable the use of time and timestamp parameters. The timezone for those reports always defaults to server-side processing – and the parameter UI correctly takes that into account. Of course URLs that have the old short dates specified are still accepted – but you may see a deprecation warning in the logs now. And last but not least, the date picker now also allows you to enter time information into the text field if your data-format pattern allows it.

The Pentaho Wiki contains the complete story on how dates should be passed around.

This entry was posted in Parameter on by .
Thomas

About Thomas

After working as all-hands guy and lead developer on Pentaho Reporting for over an decade, I have learned a thing or two about report generation, layouting and general BI practices. I have witnessed the remarkable growth of Pentaho Reporting from a small niche product to a enterprise class Business Intelligence product. This blog documents my own perspective on Pentaho Reporting's development process and our our steps towards upcoming releases.

One thought on “Its about time .. a better date parameter handling in PRD-3.7

Comments are closed.