This is the mail archive of the mailing list for the libstdc++ project.

Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: [RFC] libstdc++/9404 or weird stringbuf internals

Pétur Runólfsson wrote:


But, if someone (Pétur ;) wants to see overflow called during
a sputc which starts with pptr() == epptr()...

overflow must be called if a class derived from stringbuf is
supposed to work correctly. Google returns a number of posts
from people who have tried to derive from the standard stream
buffers (including stringbuf).

Indeed. Your report is correct... just not easy to fix ;) !

What to do, then?

I could try changing both _M_out_buf_size() and overflow
to the effect of calling overflow for each sputc of a sequence.
This has a price, however, in terms of virtual functions called.

If efficiency is a concern, sputn should be used instead of
sputc. For maximal efficiency, stringbuf shouldn't be used at all
(for example str() is an O(N) operation).

Yes, I agree with this statement of yours. In my opinion when appending single chars
with sputc the only standard conforming behaviour is calling overflow each time, even
if this has an high cost in terms of virtual function overhead.

You might also want to look at DR 169.

Yes, I read it: the only part of my patch for 9404 which is already in the final form, is
some addittional documentation inside overflow about DR 169 ;)


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]