[Bug 531252] Review Request: lfc - LCG File Catalog (LFC)

[bugzilla at redhat.com]

Please do not reply directly to this email. All additional
comments should be made in the comments box of this bug.


https://bugzilla.redhat.com/show_bug.cgi?id=531252





--- Comment #5 from Steve Traylen <steve.traylen at cern.ch>  2009-12-09 16:25:16 EDT ---

Concerning the 1 or 4 my feeling is that your method you describe of
4 * (unpacks, patch, .... build, installs) is probably better
in the long run.

With the exception of the configure stage it is a series of loops(?).
More of a pain now but better for future when you are updating the 
package.

>Since the choice of building liblcgdm together with the LFC or the DPM is
>arbitrary, the Requires should not depend on this choice, and since having
>%{version}-%{release} Requires across source RPMs is a bad idea ...

It's a bad idea normally but in this case its the same .tgz and so all
the binary packages are essentially sub-packages to the binary end user.
Certainly the two .src.rpms should/would be released together if they contain
the same upstream update anyway. Obviously doing the loop avoids the
situation.

Also the lib you are building against is not the same instance as the
one that this is then ran against. Its a private lib in the build.

I was looking for precedents of two packages starting life with the same
source.
There are 66 tar.gz, .tgz or .bz2s in more than one rawhide package so they
do exist but inspecting the list only these mostly probably for other
reasons, e.g including private libs for some reason, but there are some
that look similar to this one. e.g.
//Python-2.6.4.tar.bz2 in python-0:2.6.4-3.fc13.src
and
//Python-2.6.4.tar.bz2 in python-docs-0:2.6.4-2.fc13.src


There are certainly precedents for the multi build option, e.g one of 
mine ndoutils.

Writing this really convinces me one package is the way to go.

Lastly on a separate note rather than say
/etc/init.d/lfc-mysql 
should/could it be 
/etc/init.d/lfc 

and an ghost and alternative to lfc-mysql , -postgres instances. Would
be consistent with the other files managed under alternatives.
Is it possible to make the init.d script db independent and avoid the
situation completely?

-- 
Configure bugmail: https://bugzilla.redhat.com/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.