D18551: clang: Create preamble only on second parse

[René J.V. Bertin]

rjvbb added a comment.


  >   On my test file, time spent in clang_codeCompleteAt goes down from ~700ms
  
  Seriously, we're complaining here about something that takes 700ms the first time?? An almost 70x speed-up sounds impressive but that reeks of a synthetic benchmark. Which also included other changes.
  What I'd like to see is how much difference the user sees with or without a preamble created on first parse. If that difference is obvious it can be quantified using a screen recording and measured in frames.
  
  Evidently there should also be some better quantification (from the patch author) of what is gained by postponing the preamble creation.

REPOSITORY
  R32 KDevelop

REVISION DETAIL
  https://phabricator.kde.org/D18551

To: aaronpuchert, #kdevelop, mwolff, brauch, rjvbb
Cc: rjvbb, kdevelop-devel, glebaccon, hase, antismap, iodelay, geetamc, Pilzschaf, akshaydeo, surgenight, arrowd
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.kde.org/pipermail/kdevelop-devel/attachments/20190128/118f7c0e/attachment.html>