[kirkilj]

When we open a DITA map with hundreds of refs, it can take several minutes before the map is displayed as it parses, matches, extracts and transmits the title information across the network from the CMS and displays it in Serna. We're about to roll out the first use of Serna to our particular department, and we suspect that this might create a bad first impression.

Is there a debug trace option (other than instrumenting the code with print statements) that would allow us to examine the latencies of all the intermediate steps so we could better think about optimization opportunities? If there isn't, then perhaps this could be one of first contributions to the Open Source version. ;-)

John

[apg]

John,

I think the easiest way to speed-up the process would be to disable external reference processing in the maps. You can provide titles in topicrefs by using attribute 'navtitle' with 'locktitle=yes' in topicrefs. Otherwise for each reference Serna will read and parse referenced XML file to pull in the title; although Serna does it quickly, it may take a while if you have hundreds of references.

If you want to do a profiling, you can do so without hacking Serna code: it is possible to write a small Python plugin which will register custom XSLT function, and then you can add invocation of this function in the stylesheet around xsl:document() function which pulls in the content (look in plugins/dita/dita-xsl-serna/map2fo.xsl)