[Bug libstdc++/21334] Lack of Posix compliant thread safety in std::basic_string

["gdr at integrable-solutions dot net" <gcc-bugzilla at gcc dot gnu dot org>]

------- Additional Comments From gdr at integrable-solutions dot net  2005-05-02 18:45 -------
Subject: Re:  Lack of Posix compliant thread safety in std::basic_string

"pcarlini at suse dot de" <gcc-bugzilla@gcc.gnu.org> writes:

| ------- Additional Comments From pcarlini at suse dot de  2005-05-02 17:23 -------
| > I don't quite understand why you want to see this issue as a memory 
| > management policy issue, as opposed to a thread safety issue.
| 
| Whatever, after all that is the layer at which you actually do the low-level
| memory management. That is the level at which you decide whether actually
| freeing or allocating memory, no?

No.  You can have a thread safe or thread unsafe management.  Doing
things at low level should not imply being blind-sighted or losing
perspective. 

| But I don't care much about those names.
| 
| > I don't quite understand what "position papers" you're talking about.
| > Could you give references to them to clarify the issue?
| 
| Gaby, of course Herb Sutter papers, also reprinted in his book.

Ah, but if you said "Herb's paper", that would have rung a
bell. Saying "position paper" does not tell me much, except that he
who uses that term was being a bit condencending.

No, I'm not discussing a position paper.  I'm trying to make sure
that we do not hand wave PRs under the assumption that they would come
from "position papers". 

| > What are those subtle issues with memory management during exceptions?
| > As far as I can tell, V3 is the only one that insists on COW.  Why is
| > it that others don't have those "subtle issues"?
| 
| Because the best one, definitely have a separate mini-string of sort,
| what do you think?

You mean the trick used in VC implementation for example?

-- Gaby


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21334