Am 10/1/2012 3:25 PM, schrieb Casey Sybrandy:
Correction: I'm still seeing the slowdown. I created two test cases:
one for the fake file I showed you and one for my real data. All I do
is grab a value from the first block and return a string. Both files
are similar in structure and size (1.5 real vs. 1.1 fake). For some
reason, the real data has a pause and the fake doesn't. This is very
weird. Any thoughts on why this would be happening?FYI: I'm using the REST interface, so my code looks like this:
string postSample(Json block1, Json[] block3)
{return block1.val1.toString();}
Is there a way for me to turn on code coverage to see where the delay
may be?
You can use set DFLAGS=-profile -g before running vibe to enable
profiling, but I think you have to be lucky to get a meaningful result.
Maybe if all the strings have the same length as in the original data,
the pause can be reproduced?
Btw. while looking for something suspicious, I found and fixed a bug in
the JSON parser for unicode sequences (\uNNNN), but that shouldn't
make a real difference in execution time.