[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: RFS: libibverbs, libmthca - InfiniBand libraries



Hi Roland,

Am Donnerstag, den 15.09.2005, 09:50 -0700 schrieb Roland Dreier:
> Please let me know if there's any way I can improve the packages, or
> if more information is required.

as I am unfortunatelly not in possion of infiniband hardware, here's the
first observations:

+ zero-byte files: (can be removed, can't they?)
  * ./libibverbs-1.0-rc2/INSTALL
  * ./libibverbs-1.0-rc2/NEWS

+ both versions are "-rc2", not neccessarily a problem, but what are you
waiting for if they're now to be uploaded to Debian already?

+ Your debian/changelog has 2 lines for the closes. For my personal
preference I'd keep this in one line unless it's overly long (which is
not the case here)

+ ARGH! cdbs ;)
	(leaves stuff like:
 dpkg-shlibdeps: warning: could not find path for libibverbs.so.1)


+ your build-depends don't satisfy:
configure: error: sysfs_open_class() not found.  libibverbs requires
libsysfs.

+ your lintian version seems to be out of date:
W: libibverbs source: out-of-date-standards-version 3.6.1
W: libibverbs source: native-package-with-dash-version

+ ibverbs-examples sounds like they're not needed, yet looking at the
binaries within they're the toolbox, right? Shouldn't it be thus
ibverbs-utils or something alike?

+ no need to call the -dbg pacakge libibverbs1, the dbg should match the
exact version including SONAME of the libibverbs1 and thus can remain
libibverbs-dbg

+   SONAME      libibverbs.so.1
is a very challenging goal. Are you sure that all 1.x.y releases will
ship the same API for each and every upcoming version of x and y? Not
that this is impossible to reach, just mentioning the implications.

checking libmthca-1.0-rc2 now:
- checking for infiniband/driver.h... no
configure: error: <infiniband/driver.h> not found.  Is libibverbs
installed?
(apparently it's installed however:
$ dpkg -S driver.h
libibverbs-dev: /usr/include/infiniband/driver.h
)

...not that I really understand this one but it's fully reproducible in
my testenv. Any thoughts?

-- 
Best regards,
 Kilian



Reply to: