Re: [Boost-users] Disabling Exception Handling

Subject: Re: [Boost-users] Disabling Exception Handling
From: Noah Roberts (roberts.noah_at_[hidden])
Date: 2009-04-14 14:42:05

Emil Dotchevski wrote:
> On Tue, Apr 14, 2009 at 9:40 AM, Noah Roberts <roberts.noah_at_[hidden]> wrote:
>> Dykstra, Martin wrote:
>>> Is there a way to disable the exception handling in the Boost libraries?
>>>  Our project requires that we disable exceptions by using the -no-exceptions
>>> parameter when building (via gcc).  Is this also a valid parameter to use
>>> when building the Boost libraries?
>> I used to have to live under the same silly coding standard.  Eventually I
>> was able to convince the rest of the team, except the one guy who had the
>> power to make such decisions, that we were being really silly.  We *were*
>> linking to the standard library so any supposed benefit of not using
>> exceptions was lost (and I don't buy the supposed benefits either).
>>  Eventually that guy quit; best thing to ever happen to us actually.
> 
> The problem is that "the guy in charge" may not quit (or the platform
> you're targeting may not support exceptions) and in that case it is
> reasonable to make an effort to use STL or Boost.

It's pretty darn rare that there's actually a legitimate reason why 
exceptions won't work for you.  Of course in that situation it might be 
worthwhile to try boost and stl.  Then again, when you have a legitimate 
reason against exceptions you just might have a legitimate reason 
against templates too.  At that point you might consider if C++ is even 
the best option and you're probably doing a lot of things by hand.
> 
> The other problem is that guys in charge who don't like exceptions
> usually don't like STL or Boost either. :)

Yes, I found that true.  I was constantly having to defend myself for 
using std::string instead of a statically sized char[]!  I never 
considered myself good enough to do that correctly; problem is that he 
wasn't either but thought he was.

Like I said, best thing that ever happened to us was his quitting.  This 
guy even told me that the document/view architecture was considered bad 
by most experts now when I tried to get GUI/data model separation in the 
design.  That's the day I realized there'd be no reasoning with the guy :P

It was certainly hard to work in such a situation.  But eventually, if 
you can earn the respect of the rest of the team and the boss's boss you 
can force change for the better.  That's actually part of what happened 
to this guy ;)  Show that you can accomplish more, with less expense, 
and more stability with every little inroad you can get.  I started with 
forcing the std::string issue and slowly moved forward, pushing the 
boundary every step of the way.

Not that I'd recommend that with every superior you don't agree with. 
Sometimes (usually) they have real reasons for what they're doing.