This is the mail archive of the mailing list for the GCC 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: [PATCH] Fix PR libstdc++/19510: Uninitialized variable in someiterators

Gabriel Dos Reis wrote:

Chris Jefferson <> writes:


| >We have a default-initialisation (to zero in this case), rather than an
| >indeterminate value. IMHO any slight overhead is worth it.
| >
| Hmm.. I would tend to say the "C++" way was to never default initalise

well, that is not "C++" way.  That is something inherited from C, and
preseved there more for inertia or alleged efficiency of POD in
general than in specific cases at hand.  Here we're talking about
inline initialization of a class member.  I also see no reason to dump
annoying messages on users about which he/she can do nothing.

How do you know under the debugger that a list iterator is singular? Do you prefer to see a null pointer value or some values you cannot say off hand are valid or not?

To be honest, I perfer to initalise these things :)

However, with regards sending this back to 3.4, I'm not convinced it's an actual bug, although it is likely to lead to more informative debugging / less memory corruption (no-one should be counting on the iterator pointing at NULL of course). However if it is in fact safe to send back there as you say, I don't see why not.


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