Startup time is too long.
Let me expand, because that sounds very critical. I am using PDF::API2
because it is mature and well supported. However, I am currently using
it to extract simply meta information from PDF, notably bounding boxes
and number of pages in a document.
This run very nicely - thank you.
But I have noticed (under -d:DPro) that in fact, the majority of the
time is being spent in start up, and my actual request is dealt with
VERY fast indeed.
A brief reading of the code (guided by the profile) reveals that quite a
lot of data is read from configuration files, ready for use when
modifiying PDFs; since I don't (in fact) do that, this is not desirable.
Would it be feasible to defer reading in this data until the user
request an actual modification command?
Alternatively (since I'm in a minority) would it be possible to have a
"don't initialise data structures or import modules used for
modification" option, with would be disabled by default?