Open GoogleCodeExporter opened 9 years ago
Here's a patch to add one:
- https://github.com/mgorny/npapi-sdk/commit/9b054832e.diff
It may be also a good idea to move COPYING to the main dir, as the copyright
applies to NPAPI headers as well:
- https://github.com/mgorny/npapi-sdk/commit/1e369286c01.diff
Original comment by mgo...@gentoo.org
on 30 Aug 2011 at 4:45
Don't jab at us with phrases like "If this is supposed to be a serious project"
to imply that we're not serious if we don't do what you want. It's not
appreciated. Just make your argument and we'll make a decision.
Original comment by josh....@gmail.com
on 1 Sep 2011 at 4:25
And FYI, this is the canonical upstream header source for at least the top
three NPAPI-supporting browsers (in addition to a number of plugins; those tend
to be closed-source so it's hard to say how many). I'm pretty sure that makes
us a "serious project" whatever your personal opinion might be.
Original comment by stuart.morgan
on 2 Sep 2011 at 9:17
If this is a canonical header source, it would be nice to have a
canonical/supported/common way of installing and compiling against the API.
Providing a buildsystem and pkg-config support would provide a common and easy
way for package to discover and compile against NPAPI. This would make things
much cleaner for distros.
Please consider distributing such a buildsystem with NPAPI.
Original comment by ohnobinki
on 2 Sep 2011 at 4:43
Here are my patches which:
1) create a simple autotools install for it (installing headers into
/usr/include/npapi-sdk),
2) move COPYING into top-level dir (so that it is clear that LICENSE applies to
all files within),
3) generate NPAPI version consts from configure.ac (it is nicer to have the
version defined on one place only, and configure.ac needs one const).
The third patch is based on sources with pkg-config file added but can be
rebased easily to work without it.
Original comment by mgo...@gentoo.org
on 4 Sep 2011 at 10:12
Attachments:
Requiring everyone to run configure would make this much more painful for Mac
and Windows developers (as compared to just having to check out the source and
go). I'm all for making this easy to use on Linux, but not at the expense of
other platforms.
Original comment by stuart.morgan
on 5 Sep 2011 at 5:39
Then please just ignore the last patch here. This will make the package fully
compatible with autoconf-less run. Windows/Mac users could just grab files from
the package/svn as they do now, and Linux users could use configure to get the
pkg-config file as well.
Original comment by mgo...@gentoo.org
on 5 Sep 2011 at 6:47
Well, ignoring the last patch shouldn't be necessary. The tarball generated by
`make dist' includes a working npapi.h, so packages should be no less usable
than before on both Windows and Mac OSX as there is no need to run ./configure.
From what I remember, Mac OSX should be able to run the ./configure shipped in
a dist tarball out of the box. If you happen to have Xcode installed, as most
people developing on a Mac OSX box would, you also get a working make which
would allow you to install npapi-sdk to a prefix. Xcode also comes with a
working `autoreconf -vfi' which works with a checked out repo. (OK, I'm
confused; how are Mac OSX developers supposed to check out npapi-sdk's repo?
Oh, apparently Xcode includes SVN so we are already assuming that the Mac OSX
developer has XCode...)
The guides I've seen for compiling things on Windows point people to using
precompiled binaries for dependent libraries and such. Using a dist tarball to
get the npapi.h wouldn't be that different...
Original comment by ohnobinki
on 5 Sep 2011 at 2:01
I wouldn't add any build system here. Even WebKit uses 8(!!!) build systems
http://trac.webkit.org/wiki/Unifying%20the%20build%20system
I wonder what other build systems might be used in plugins.
These sources can't be used as is in browsers. Each browser has its own copy
checked in the source control.
Of cource, adding a build system might be harmless, but it might be useless as
well.
Original comment by Van...@gmail.com
on 23 Feb 2012 at 10:28
Original issue reported on code.google.com by
mgo...@gentoo.org
on 30 Aug 2011 at 4:21