Subject: Re: [boost] [system][filesystem v3]Question about error_codearguments
From: Domagoj Saric (dsaritz_at_[hidden])
Date: 2009-11-07 12:05:14
"Stewart, Robert" <Robert.Stewart_at_[hidden]> wrote in message
> [snip much too much text to read, let alone study]
> I imagine there was useful information in your, but I won't be
> commenting on it as it stands, because I can't afford to take the time
> it would require. You might try sending a more succinct version of that
oh well :)
perhaps i can give you a short introduction/"a bit extended toc"...
firstly i tried to fix the ambiguity of my first post by separating the two
problems of checked error codes and of the error-codes vs exceptions debate...
then, after briefly summarizing and 'solving' the first problem i identified
the "three schools of thought" in the error-codes vs exceptions debate...
...there i expanded on the arguments from the boost::error_code documentation
in favour of the hybrid approach and against either of the "fundamentalist"
approaches (only exceptions or only error codes)...
then i summarized the three hybrid solutions so far proposed in this thread
(error codes, overloads and "smart error objects") trying to document their so
far percieved advantages and disadvantages...
then i acknowledged that a hybrid approach is not universally applicable and
tried to identify and define the line when a switch to exceptions must be
ps. the probably largest part, in the middle of the post, is an "efficency
rant" with real life examples of the effect that the presence of exceptions has
on generated code as well as a 'case against' the 'managed' trends in c++...in
general your garden variety "i do not want to pay for what i do not use" cry
-- "That men do not learn very much from the lessons of history is the most important of all the lessons of history." Aldous Huxley
Boost list run by bdawes at acm.org, gregod at cs.rpi.edu, cpdaniel at pacbell.net, john at johnmaddock.co.uk