[Klone-users] fatal source errors ignored?

thomas fossati tho at koanlogic.com
Sat Aug 4 14:04:45 EDT 2007


Hi Mark,

On Aug 4, 2007, at 6:57 PM, Mark Richards wrote:
> Hope your weekend is sunny and fun.

too hot :)

> I added --enable_warns so it looks like this:
>
> 	KLONE_CONF_ARGS ?= --disable_cxx --enable_warns
>
> Then I created a problem by doing this in index.kl1:
>
> 	// missing parameter
> 	snprintf(cSql,1024,"/runtime/exec/senddata.sh %i /var/tmp/comms.OK 
> comms.OK inbox/");
>
> kloned still builds.  There are a bunch of warnings but the build 
> process just keeps on rolling.  I only see them if I scroll way back 
> in the make listing.

if you want the build process to stop on warning you have to tell gcc 
to '-Werror' but that's a little bit strong, and moreover it breaks the 
klone build as well - since we have a few harmless warns still laying 
around :)

Anyway if you want to play with the '--enable_warns' CFLAGS to pick up 
those '-W...' that you need you'll have to patch the klone core 
configure (see __makl_enable_warns() function), perhaps adding your own 
warn function and flags.

> When this happens I noted that the warnings which follow the 
> legitimate issue (too few arguments for format) are really not valid 
> warnings at all.  It's almost as if the compile process is thrown off 
> somehow.
>
> Would be ideal if there were an option for halt on warning or if the 
> make process, if there are warnings, tells you at the end (lists them 
> out).
> ...
> Since investigating this I find that there are a bunch of potentially 
> serious warnings way back in the make listing that I should have dealt 
> with.  So again, would be great if somehow the process told you about 
> these at the end, or maybe just kept quiet until something breaks?

you mean something like:

$ make > make.out 2> make.err ; less make.err ?

ciao, t.

PS perhaps you'd appreciate pretty_make.py ( 
http://phil.freehackers.org/pretty-make/index.html ), personally I like 
it very much :)




More information about the klone-users mailing list