At 10:26 AM 31/01/2004 +1100, Mark O'Donohue wrote:
>Helen Borrie wrote:
>>Hello Nickolay, Mark and whoever,
>>I clean-installed RC8 classic from the rpm onto Red Hat 8. I'm in the
>>process of downloading the SS version, which I'll try to clean-install
>>onto Mandrake 9.0.
>>Now, experience so far with the classic install is not fantastic. In
>>short, it doesn't do a lot of the things that Mark's new notes (please
>>read release notes v. 107) say it should.
>
> > There is no firebird user, no firebird group
>
>I'll check this one, there was a bug in RC7 where classic where the check
>to see if fb user already existed always returned true, (or was that fb
>group?) but it was fixed.
>
>and a remote SYSDBA client cannot access any databases
>>except /opt/firebird/examples/employee.fdb.
>>e.g.
>>ISC ERROR CODE:335544352
>>no permission for read-write access to database /data/sample/lstore.fdb
>
>
>For remote access, your database has to at least be in the group firebird
>so the server has permission to access it.
>
>You will need to do:
>
>chgrp firebird /data/sample/lstore.fdb
>
>(Im pretty sure the above is enough, but possibly you might need
>chown firebird /data/sample/lstore.fdb as well)
This isn't documented. In any case, if there is no firebird group and no
firebird user then it isn't this simple (even if people *were* aware that
ownership was the problem...and it's not obvious.)
>>I think this should possibly reflect the case if DatabaseAccess were set
>>to Restrict but, according to the firebird.conf file that was installed,
>>the default is Full.
>>In addition, this RC went out with an old (broken) version of
>>changeDBAPassword.sh.
>
>There was a bug, (it produces warnings, but still works) it was fixed a
>day or so after CS RC8 was released, so the fix made it into SS RC8.
Sure, but people are still downloading CS.
>>Others questioning in support have this problem with RC8 classic as
>>well. So, three things:
>>1. When can we expect a RC installer for 1.5 Classic that actually works
>>properly?
>
>I'll check Helen, but as far as I know it currently does work.
It works, inasmuch as it completes without errors. But it leaves the
permissions in a not-as-documented state - so it's a problem that requires
fixing and proper testing to ensure it *does* work as documented.
>>2. How is anyone supposed to know what it's meant to be like when it
>>*does* work properly?
>
>Helen, if you have issues, as above raise them,
..as I have...
>but the installers have been basically the same since fb1.0. The only
>real change for fb1.5 is that the default run user is not root.
And it's a real change. It makes the difference between having an
installer that "just works" (as previous ones did) and one that only "sort
of works".
>But the way you write it it sounds like the sky is falling in, because an
>odd link is incorrect etc.
To someone who is around the OS every day, for years, the "odd incorrect
link" is not a big deal. You know how to fix it. "The sky is falling in"
is a gross over-exaggeration. But taking a "near enough is good enough"
attitude to the product's human interface is not destined to win hearts, is
it? Combine that with the confusing and ugly download interface that
confronts the newcomer and you've got a fairly strong potential to turn an
otherwise good open source product into an Aunt Sally. We owe our
developers and fellow project workers something better.
>>3. Who is going to take responsibility for ensuring that the release
>>notes are correct with respect to the Linux installs? (beyond saying
>>that they are wrong, without providing any details or corrections...)
>
>Helen, I've done one review and they, as far as I know *are* in sync, (or
>at least very close).
"Very close" is not close enough, IMO. It's spoiling the ship for a
ha'p'orth of tar.
Helen
|