I'm new to sed, I have a string
lightdm --session-child 14 21
the two numbers can change and I would remove everything after lightdm
I tried various way but none worked
sed 's/ --session-child*//'
sed 's/\ --session-child*//'
sed 's/ --session-child*$//'
and others, but none worked where is wrong ?
* in regular expressions means "repeat the previous character (or character class) arbitrary many times. So, --session-child* matches
--session-chil --session-child --session-childd --session-childdd --session-childdddHowever, it does not match --session-child 1, because a is not a d! (and neither is a 1, but that doesn't matter at that point anymore. The streak of d is broken.)
If you want to say "any single character", that's a . in regular expressions. If you mean "arbitrarily many arbitrary characters", that's .*.
So, you'd need sed 's/ --session-child.*//'
However, I tend to advocate for matching the things you want instead of erasing the things you don't want. I don't know what you originally intended, but assuming you wanted to preserve anything before the first space, I'd write that as sed 's/^\([^ ]*\) .*$/\1/'. Meaning: from the beginning of the line (^), start a new group (encapsulated in \( … \)). That group contains any character that is in the character class (enclosed in […]) described by ^ , which means "everything but" (^) the space character . Repeat that as many times (0 or more) if possible (the first *).
If a space comes after that, absorb that space and all characters until line's end ($) and replace them with the first (and only) matched group, \1.
That's way more effort to write down. Why would I do that? Because writing down what you not want is often a mental exercise that involves assumptions that are not inherent to the actual problem. For example, here you assume that the first argument to whatever comes before is --session-child. Now, if that "whatever" isn't lightdm, why would it take a --session-child argument? And if it always was lightdm why, would you use sed at all? You would just output lightdm. So, by not making assumptions on a "cute little side fact" but focusing on what I really care about, I can avoid future bugs with input I didn't expect.
sed 's/ --session-child.*//' works. I will study sed 's/^\([^ ]*\) .*$/\1/' to better understand it
awk instead might be easier: awk '{print $1}'
*matches zero or more characters) with regular expressions (where.*matches zero or more characters) - see for example Why does my regular expression work in X but not in Y?