[
Date Prev][
Date Next][
Thread Prev][
Thread Next][
Date Index][
Thread Index]
[
List Home]
|
RE: [cdt-dev] CDT parsing & indexing
|
Title: CDT parsing & indexing
Hello,
It
seems that parser is called everytime say code completion is invoked.
This
would mean searching for all the includes one by one in all the directories.
Thus, having large no of directories affect code completion time
everytime. Right ?
What I have noticed is Code navigation per say was fine
but coming back to the file from which I navigated froze the GUI for some
time.
Also,
Outline was slower with large no of directories.
Also,
would like to know if "No Indexing" is on and code completion is invoked, does
parser goes to the file system to locate files or it ignores the action ?
/ Veenu
Hi,
to
find the header files included with include directives (#include) the parser has
to check one directory after the other until the file is found. System include
directories are checked last, such that the parser has to go through all of the
user-specific ones first.
So using 650 include directories is in many cases 65*
slower than using 10 directories.
Yes, the number of include directories can have an
impact on the speed of the indexer. However, after the index is built, it should
not impact the navigation feature.
Markus.
From:
cdt-dev-bounces@xxxxxxxxxxx [mailto:cdt-dev-bounces@xxxxxxxxxxx] On Behalf Of
Veenu Verma
Sent: Freitag, 28. September 2007 08:52
To:
cdt-dev@xxxxxxxxxxx
Subject: [cdt-dev] CDT parsing &
indexing
Importance: Low
Hello,
Need some help in
knowing the functioning of CDT parser
A little background first...
We saw the slow editor problem(reported by
Christopher on the forum earlier) and thankfully it is resolved by moving to
CDT4.0.1
But indexer was still slow for us (2-3 minutes to get code
completion) until we made some changes
Initially number of directories included
using "Include Paths" option was extremely large (like 650 as we have huge no
of interfaces) Indexer, code navigation etc became really slow. With CDT4.0+,
indexer took 1 hour to built in comparison to old CDT when it took couple of
minutes and On returning to the file after code navigation, GUI was
freezing for some time. Also code completion was taking quite
long.
Reducing the no of directories
though keeping the number of files same as before (around 3000) did the trick
for us.
And code completion, indexing and code navigation have become quite
fast now.
Just curious to know how does parser works with traversing directories
included with "Include Paths" option ?
Also, How does it affect indexer and
navigator ?
Is
there any recommended limit for Include path directories ?
PS: We are using Eclipse europa
with CDT4.0.1 now
Veenu Khanna
