[uClibc] uClibc++ 0.1.11 is out!
Garrett Kajmowicz
gkajmowi at tbaytel.net
Sat Jul 2 18:19:08 UTC 2005
On July 2, 2005 01:09 pm, Mike Frysinger wrote:
> On Saturday 02 July 2005 10:49 am, Garrett Kajmowicz wrote:
> > Finally, after a lot of work, I've decided to put out a new version of
> > uClibc++.
>
> i just built & ran `make test` on x86_64 and all but one test passed ... it
> looks like just a bug in handling of pointers ? maybe assuming they're
> 32bits in size ?
The test suite assumes that long ints are 32 bits in size. The code works
properly, however. The spec states that there is a type (usually unsigned)
which represents the size of items - size_type. The invalid value (also the
largest possible value a container can hold) is generally done as (unsigned
long int)(-1). Yes, that assumes two's compliment math and makes the code
unportable to machines which nobody else uses (correct me if I am wrong).
However, if you look at the actual test output, you will notice that you have
lines which state:
Position of "q": 1844674407370955161 - should be 1844674407370955161
Notice that the two numbers match. In short, the code works properly but the
canned output is for 32bit machines. I might work on this in the future, but
this isn't a high priority.
- Garrett
More information about the uClibc
mailing list