I have narrowed down the major performance regressions to be code where evalonprint=true is used. Testing on a 300 page PDF with a header and footer, the PDF is generated in a matter of seconds. Adding *evalonprint * however makes it run forever without finishing.
Here is a test showing the terrible performance of evalonprint. Changing evalAtPrint to false will let this code run in seconds.
what is the time difference in ACF with and without the evalatprint?
In ACF "with and without evalAtprint", it took the same time( 15 or 16 sec)
You are correct that feature was not implemented prior, but I still had that code in my code base since Adobe implements it, so as soon as I upgraded, my code had a performance regression. In other words, the same CF code had no performance issue on 5.3.4 and had a performance issue on 5.3.5. Whether it's the addition of a feature or the modification of a feature, it's a regression when my code stays the same and a Lucee upgrade breaks it.
In this case, the temporary fix for me is easy – just to remove the evalonprint attribute, however someone with a codebase running on Adobe CF who needs that attribute cannot do that.
i assumed that "evalatprint" was not simply ignored, i assumed it was throwing an exception like "attribute evalatprint is not supported", so in that case i agree, it is kind of a regression, i will add the mark again.
traded out of sprint in favor of LDEV-2876