I run the diff from the comment parent directory of the two trees I wish to diff. The .svn parameter is the name of subdirectories I do not want included in the diff. (I use subversion, if you hadn’t guessed.) The -p1, on the patch command, causes the first component of the path (olddir/newdir) to be ignored.
In addition to all the normal reasons why unrestricted TLDs are a bad idea, there is one additional downside:
Unlimited TLDs will make unfamiliar configuration files more difficult to understand.
If you are looking through pages of configuration and see “RandomProperty=10.0.0.5” you know it’s an IP (v4) address.
If you see “RandomProperty=foobar.com” you know it’s a domain name.
If you see “RandomProperty=foobar”, what is it? It could be a new TLD, it could be an unqualified domain name, it could be a directory/file name, it could be anything at all.
Hopefully ICANN will charge £1m for the new TLDs, so their use will be minimal. If you can get a TLD for £1k, thousands will proliferate. It’s rather strange how the UK media has covered this topic, drawing parallels with ‘opening up the wild west of America’. Perhaps they should draw parallels with allowing people to choose their own telephone numbers; where all information on what kind of number it was, would be lost (mobile/landline/geographic region/free/local rate/etc.).
Sometimes it’s useful to be able to convert a .a file into a .so shared object library. A better way is to modify the Makefile, or dig deep into autotools, and recompile/relink.
The command I use is:
ld --whole-archive-shared-o lib.so lib.a
Where ‘lib’ should be replaced with the name of the library which you are converting. This has worked for me, but your milage may vary.
The created shared object library will be larger than a properly created .so. This is because the object files in the .a library will have statically included some libc (and other) code which a properly created shared object library could share instead.