Re: [boost] Formal Review of Proposed Boost.Process library endsSunday!

Subject: Re: [boost] Formal Review of Proposed Boost.Process library endsSunday!
From: Domagoj Saric (domagoj.saric_at_[hidden])
Date: 2011-02-21 06:48:51

Sorry for being a day late...

> - How much effort did you put into your evaluation? A glance? A quick 
> reading? In-depth study?

As I do not yet need the functionality of the library but might need it in 
the future for simple cross platform (Windows and OS X) executing of other 
existing binaries/child processes (with perhaps adjustment of the arguments 
passed and of the child process priority) I took a simple glance over the 
library code (the one in the sandbox) to see whether it 'goes in an 
acceptable direction':

> - What is your evaluation of the implementation?

And I immediately found material for my typical rant against a practice too 
often encountered in many Boost libraries (e.g. a related one 
http://lists.boost.org/Archives/boost/2010/10/172390.php):
the implementation is inefficient to the point of being plain evil: 
compiling a for-size optimized MSVC++ 10 x86 release build of the 
start_child.cpp example (that simply starts a child process) produced a ~100 
kB binary...!?
"Why on Earth" must one pay for strings, vectors and maps of strings, shared 
pointers, streams (!?), exceptions and all the other 'goodies' and complex 
logic only to (after 'getting your head dizzy' by stepping through the code) 
finally arrive to the CreateProcess() call...?

Sorry for the 'rant', I appreciate any effort for public contribution, but 
my vote, if it still counts, being a day late, is NO...

-- 
"What Huxley teaches is that in the age of advanced technology, spiritual
devastation is more likely to come from an enemy with a smiling face than
from one whose countenance exudes suspicion and hate."
Neil Postman