Compiler cache

[Victor Röder]

Hi!

The Linux Magazin (german) writes in its latest issue (09/2001) an article 
about a 'simple' compiler cache for C/C++. It's based on the assumption that 
source files don't have to be recompiled if there was a change in a comment 
or something comparable. Given is the following situation: There are 50 
header files and 50 source files. Now a wrong word in the documentation of 
one header file has been corrected. Unfortunately 45 source files depend on 
this one header file. Therefore all these files have to be recompiled 
although the result will be the same. And another situation is mentioned: A 
project build switches often from debug code (-g) to optimizied code (-O2). 
Everytime a 'make clean' and 'make' with changed compiler options is needed 
though the compiler has compiled them in the past already. Other advantages 
are time improvments while packing RPMs and shared development via CVS.
In the article two disadvantages are mentioned: Speed and disc space. Speed 
because the compiler cache does complex operations on the source files 
while 'make' only combares the modification times. And disc space because 
there is a cache :-) and with larger project this cache could become large (I 
don't know how large :-)
The compiler cache is according to the article not ment to replace make but 
to combine these two tools.

Note: I'm neither the author of the article mentioned nor the author of the 
compiler cache. I have read this article and thought "Might be good for 
Gideon!". And I don't want such replies like "Yes great idea! Just implement 
it!" I just want to read your opinions and questions about the compiler cache.
I haven't a idea already how long the compiler cache takes to find out if the 
recent file has to be compiled or not. I have to inform my self :-)
BTW.: The URL is http://www.erikyyy.dercompilercache/ for interested people.
In addition for german people interested: Buy the new Linux Magazin ;-)

Bye,
	Victor


-
to unsubscribe from this list send an email to kdevelop-devel-request at kdevelop.org with the following body:
unsubscribe »your-email-address«