Sebastian Redl skrev:

Thorsten Ottosen wrote:

...

Hi,

In code like this:

std::string columns() const { return (boost::format("%7s, %7s, %7s, %7s") % "testCost" % "optimalEcr" % "efficiecyEcr" % "testCostEcr").str(); }

it is very annoying to have to put () around the whole expression and then call .str() (Just imagine that we are creating a string on the fly to a function).

Therefore I suggest that we add

operator string_type() const { return str(); }

to the interface of basic_format<>.

Any comments? You could write it as

return boost::str(boost::format("%7s, %7s, %7s, %7s") % "testCost" % "optimalEcr" % "efficiecyEcr" % "testCostEcr");

yes, which is hardly any better than the above. Why is such a function even there? I my mind formatting is so closely tied to strings that I don't see any problem in the implicit conversion. IIRC, String.Format in other langauages return a String. -Thorsten

Alex MDC wrote:

...

Therefore I suggest that we add

operator string_type() const { return str(); }

to the interface of basic_format<>.

Any comments?

Sorry to dig this up, but I am +10 for this change.

I'm currently in the process of removing hundreds of old fixed buffer + sprintf() style log messages from our code and this would make my life a *lot* easier (not to mention more readable).

Hi Alex, Why do you prefer boost.format to sprintf? I gave up on boost.format when I realised that, like std::operator<< and boost::lexical_cast, it formats a uint8_t as a char even when I use %d. I now use something like this to get a std::string using sprintf syntax: #include <string> #include <cstdarg> #include <cstdio> #include <malloc.h> std::string format(const char* fmt,...) __attribute__ ((format (printf, 1, 2))); std::string format(const char* fmt,...) { va_list args; va_start(args,fmt); char* p; int r = vasprintf(&p,fmt,args); if (r<0) { throw "vasprintf failed"; } va_end(args); std::string s(p); free(p); return s; } Of course this lacks the ability to output types that define their own operator<<; the only one that I really miss is std::string. Maybe this can be fixed when we have variadic templates. Phil.

for me also +10 for the change. maybe it should be considered to make work also %d for a uint8_t...

Hansjörg

Add my +10. B/Rgds Max