At AMPATH we have been in process of converting our forms from using Form Entry to using HTML Form Entry module. We expected to have faster loading times as html forms are expected to be light weight as compared to files used to render forms in infopath. Unfortunately that is not the case in our initial testing on our production server. We have forms that take between 30 to 90 seconds on LAN. It has been noted though the timings may vary from time to time.
Chrome speed profiler indicated most time is spent waiting for content than time taken to transfer between the server and client. YourKit java profiler didn’t show any method taking longer than necessary. I would like to know if there is anyone who has had this experience and what was done to fix it.
@darius, @mogoodrich, @ball I hope you can share something regarding this issue.
@willa, a thing about HTML Forms is that they aren’t pre-compiled like the infopath ones. However 30-90 seconds is a very long time.
How big is the form in question? How many references to concepts does it have (approximately)? Can you share the html form definition?
I haven’t counted all of the concepts, but to give you a ballpark…there are around 280 observations (obviously each with an obs conceptId plus 1 or several answerConceptIds). If it is helpful, the form is here. Thanks!
Thanks Sara. Maybe @ball can comment on how that compares to the size of
our larger forms.
@willa, can you share the yourkit output somehow? At a minimum it would be
helpful to know where it says the hotspots are, but the full snapshot could
be even more helpful.
@sarmson @darius @willa
Quickly looked at one of our longer forms (Cardiology Consultation; probably not the longest). It has 392 obs. Loads in about 4 seconds. Running OpenMRS 1.9.
Partners In Health
What versions of core and Html Form Entry are you running?
We are running 2.6-SNAPSHOT. We wanted to use the latest changes I made to make drugAnswer attribute of the obs tag work.
Perhaps the performance degradation is related to this change. A missing index?
Strange. Seems unlikely that that change would cause as issue, but I guess it’s worth testing.
What version of core are you running at Ampath again?
@mogoodrich we are using OpenMRS core 1.9.7