castor/utils/UtilsTest.

objects representing standard in and standard error.

By being able to provide dummy output stream objects, this modification allows
C++ daemon objects to be instantiated in unit tests without the worry of them
writing to the real standard in and standard error streams.

    castor::server::Daemon

xroot clients. To migrate the prestage test we'll have to get xroot
4.0.0.

[canoc3@lxc2dev3d2 CASTOR]$ grep -rl DbSegmentCnv .
./nbproject/configurations.xml
./nbproject/private/configurations.xml
./castor/db/cnv/DbSegmentCnv.cpp
./castor/db/cnv/DbSegmentCnv.hpp
./.git/index
./.git/objects/pack/pack-a749e3127a204825b5c252b2db64ec64d9c957e8.pack

I have commented out the offending log statement because it logs
st_size modulus 4GB.  The log statement is therefore more
confusing than useful.  If anybody wishes to fix the following
log line then they MUST test their fix by migrating a file
larger than 4GB and see the correct value logged.

Please note that the log statement was never used in production before I
modified a few days ago.  This was because the log was set at DEBUG level.
I now regret moving it to INFO level.  Right now this log is NOT priority.

Fixed in the v2_1_14Version branch.

        This flag is set when doing a stat with xrdfs. For this to work we needed to
        set the inode value to 0.

        and it gives priority to individual mappings rather than using directly the wildcard "*"

Added constraining reads to avoid boilerplate in the user's code.
For example, when the file format expects to find a file mark, just calling
driveGeneric::readFileMark() will implicitely check everything.
Created corresponding exception classes.
Also changed the data blocks signatures to the more classis void *.

Fixed the read session memory management layout as the file size if not know before hand
(and not really needed, as we read files and consume free blocks in sequence.
Extended the description of the file support classes.
Fixed minor compilation issue.

This reverts commit 5bdc2b33.

This is in fact the revert of the revert (now under which cup is the coin?).

I am following the suggestion of Giuseppe that in order not to lose Eric's
work it would be a good idea to put his fix for his RFE into the master branch
as soon as possible.

enable/disable protocols.
Also extended a test case to check for bogus protocols in the requests.

to get a consistent view between files and segments, given that
the two queries are run several minutes apart.