Re: [Vala] Questions about vapi generation and testing



On Wed, 2011-10-05 at 23:09 -0300, Damián Nohales wrote:

El 05/10/11 22:07, bsquared escribió:
Luca Bruno<lethalman88 gmail com>  writes:

On Wed, Oct 5, 2011 at 6:30 PM, bsquared<bwcode4u-Re5JQEeQqe8AvxtiuMwx3w public gmane org>  wrote:

2. What can I do to optimize the generated c code in my tests?
   The generated code is 4x the size of original test code
   (test-couchdb-glib.c) as is the executable.


For the C code, nothing. About the executable, optimize with -O3 (or -Os).

I was thinking there may be some kind of best practice to minimize the
number of temp vars in the generated code.

No, I don't think so, those temp variables appear because the manner in 
which the Vala generate the code, the way to generate less code is to 
modify the Vala compiler but not to optimizing your Vapi.

You should not need to expect that Vala compiler generates a C code as 
if you had written by hand as you should not expect the C compiler do 
the same with the assembly code generated.


While I agree that the temp variables have no impact on performance
(they should be optimized out at -O1 not mentioning -O2 and -Os) they do
have on debugging. Scrolling through list of ~60-70 variables named
_tmpXYZ_ in backtrack so the do have some impact.

I guess in long-term it would be beneficial to have Vala support in gdb
but right now we don't.

Regards

Attachment: signature.asc
Description: This is a digitally signed message part



[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]