Re: [libvirt] [PATCH 1/4] Turn virSecurityManager into a virObjectLockable
On 02/19/2013 04:16 AM, Michal Privoznik wrote:
It seems like you squashed in changes made to HACKING in. I believe
that
was by mistake. However, if HACKING gets re-generated on my system,
git-diff complains about it.
I've noticed that the version of xsltproc in RHEL 6 and Fedora 17
(libxslt-1.1.26) generates different contents for HACKING than the one
in Fedora 18 (libxslt-1.1.27); I don't know if this is an unintentional
regression in upstream libxslt, or what we could do to work around that.
One thing I am wondering about, HACKING is not getting re-generated even
if I make a change to one of its dependency. Is it because HACKING
itself isn't a dependency or in .PHONY?
That I'm not sure about - HACKING is supposed to be regenerated any time
the timestamp of docs/hacking.html.in changes.
The other thing - should we completely drop HACKING from git and
generate it with other dist hooks, like gen-AUTHORS or gen-Changelog?
Do we want HACKING tracked by git?
Having HACKING tracked by git is useful if only so that people doing a
fresh checkout can read it. If you are unfamiliar with the source, then
you are more likely to look in the ALL_CAPS files in the top directory
than you are to find docs/hacking.html.in to learn about how to build
the source. In other words, HACKING is one of the few generated files
where we make an exception and include it in git, because it is needed
to tell the new user how to create all the other generated files that
would normally be present in a tarball.
--
Eric Blake eblake redhat com +1-919-301-3266
Libvirt virtualization library http://libvirt.org
Attachments:
- signature.asc (application/pgp-signature — 621 bytes)