*MY* documentation is for the released version, which doesn't have a changing directory structure. The docs will be updated as appropriate once the 0.17 release is out.
If multiple people write in to the mailing lists with problems that are caused by the ebuild in question, and wouldn't have happened if they were using a straight CVS checkout, what exactly am I supposed to think? That the ebuild's working perfectly?
Well, arguably if he build from cvs without an ebuild and followed the old instructions then he would have got the same error as well? Probably it would have been easier to work around by just trying to make again though...? However, he didn't *seem* to be pointing the finger to me, and the post had all the relevant debugging detail in it, so it was pretty obvious what his problem was...
...I'm a gentoo fan, so don't mind my opinion. However, I'm not really sure why everyone is getting so hot under the collar here?
At the end of the day, the ebuilds are just a way of passing around a script to build your software. People who build RPMs I'm sure have to apply a few patches to target them for the respective distro (eg Redhat puts some files in funny places, etc, etc). ebuilds can be thought of as a way to actually make the RPM build scripts public (in a way).
The weakness of this is of course making sure that your script is up to date with the software version. It doesn't just happen when following cvs. For example when a new version of XXX is released then usually you can just bump the build version number and install the package (can be useful if you like to live on the bleeding edge). However, sometimes something small will change on the build script, and so the ebuild needs minor tweaking before it can be released with the new version number - however, do remember that usually no tweaking is required and usually only minor changes are requred to a build script between versions (arguably no build script should be required for software at all other than "./configure && make && make install", but this doesn't seem to be the case in practice...)
So if the weakness is keeping the script up to date then the only way to fix this is to find a better way to keep the build script fresh. It's not like it needs changing every week, I think it has only needed two changes in recent memory for libsoundtouch and libmpeg2? Personally I update most of my myth modules using some ebuild cvs scripts that I wrote for version 0.09 - they are still working fine now...
I really don't see why we have had so much fuss about such a little issue? The way I read the original post was a really neutral post asking for help and including a really sensible amount of debugging info. If everyone else had said nothing except to direct him to the latest update of the script then we wouldn't have wasted more than two posts and the guy would have had a fully packaged update to latest cvs (avoids a load of problems with old versions of lib files, which has bitten me a few times...)
Can we not accomodate a couple of posts about updating a build script a couple of times a year? Is that so much to ask?
Ed W _______________________________________________ mythtv-dev mailing list [email protected] http://mythtv.org/cgi-bin/mailman/listinfo/mythtv-dev
