Red Hat Bugzilla – Bug 193559
my_print_defaults needed in main mysql package
Last modified: 2013-07-02 23:09:52 EDT
mysqldumpslow, which is in the main mysql package, needs my_print_defaults to work. Unfortunately my_print_defaults is in the -server subpackage. It would be nice if my_print_defaults would be moved to the main package; I'd rather not install a DB server just in order to examine the slow queries.
I suppose that dependency didn't exist when the packaging decisions were first made. I'll see what I can do, but if my_print_defaults adds a lot of dependencies of its own, the move may have to be in the other direction :-(
Sorry about not mentioning it initially, but I already checked the dependencies (ldd output only) and it looks to me like there would be no additional ones.
I looked at this again and am unsure what to do. It seems to me that mysqldumpslow is useless except on a server machine, and therefore it ought to be moved to the -server package. On the other hand, my_print_defaults appears to be functionality that would be useful for scripts even in a client-only installation. I checked MySQL AB's own RPM specfile, and they too have got mysqldumpslow in the client package and my_print_defaults in the server package, so they haven't thought it through either :-( I'm a bit hesitant to move my_print_defaults to the base package because of bloat concerns; the executable is a megabyte at the moment. (I think this is an upstream bug caused by careless library inclusion, but I dunno if they intend to fix it.) So right now I'm leaning to the mysqldumpslow-to-the-server-package answer. Comments? In particular, what's your use-case for running mysqldumpslow on a client-only machine?
Use case: I'm not running anything that is not absolutely necessary on a production database server, but rather copy the slow query log elsewhere for examination. Having to install the DB server package in that "elsewhere" solely for this purpose is overkill. Anyway, moving mysqldumpslow to the server package would be an improvement over the current situation, but I'd personally be inclined to do it the other way around.
I find that argument unconvincing --- if you're doing server management work, it's not unreasonable to expect you to install the server RPM. I think a considerably stronger argument could be made that my_print_defaults belongs in the base RPM on its own merits, ie, it'd be useful in other scripts besides this one. But I'm not sure what to do. I've filed an upstream bug at http://bugs.mysql.com/bug.php?id=20216 to see if they have any opinions about it.
Well, the upstream decision was to move mysqldumpslow to the server package, so that's what I've done.