Hello everybody
I am reading the e-book "the Linux command line " by William Shotts Jr.
In page 309, he writes :
" since sed, by default, accepts only basic regular expressions, several of the characters in
our regular expression will be taken as literals, rather than as metacharacters. We can
solve both these problems with a liberal application of backslashes to escape the
offending characters:
sed 's/\([0-9]\{2\}\)\/\([0-9]\{2\}\)\/\([0-9]\{4\}\)$/\3-\1-\2/' distros.txt "
Now there are 2 things I cannot understand
1) how are the backslashes executed as escape characters since they are within single quotes ?
2) How does the application of backslashes solve the problem that sed doesnot accept ERE ?
Any help would really be appreciable

1. I may be wrong, but single/double quoting only affects the possibility of variable substitution, NOT character escaping.

The single quotes tell your shell that it should not process any escape sequences in the string. If you replace sed with e.g echo, you can see what sed actually receives as a parameter.

Sed what?!

GNU sed supports POSIX BREs fine; \([0-9]\{2\}\)\/\([0-9]\{2\}\)\/\([0-9]\{4\}\)$ is a perfectly ordinary basic regular expression.

If you meant ERE, then the GNU sed manual tells why in the Extended regexps chapter: EREs and BREs differ only in that certain characters must be escaped. In other words, if you escape all ?, +, (, {, ) and } characters in an ERE, you get a BRE.

Thanks, I meant ERE indeed , I will correct it in my post . I didnt know that escaping converts these ERE metacharacters in BRE metacharacters.
I am completely covered and will mark the thread as solved

You're welcome. Technically, there are some additional differences between POSIX BREs and EREs, covered well in the Wikipedia POSIX Extended Regular Expressions article, but they do not apply to GNU sed.