I found this quite an interesting thought.
">>"
if of course the append operator, it will stick STDIN onto a file of your choice, so it gets added on the end.
Great! Everybody finds that useful.

But how about prepending?

I mean, sending some STDIN to the beginning of a textfile.
There seems to be a "forward" view to memory. It's OK to add and append stuff, but going backward, seems less conventional.

So, I thought about dong a little C program for this. But maybe this han been come across already. On first glance it may appear easy, but it may turn out in practice not to be.

Ways:
One:
That's disgusting. IN perl, you could slurp the oldfile.txt and then print out, with the "new first line" coming first. I must admit, I would do the same with the C program. Read in the oldfile.txt first.

append to tac'd version of the file. Then re-tac while overwriting file.

Nope. too messy.

I'm sure there are probably a dozen ways to do what you want. I know sed can insert lines at the beginning of the text, for example. But even just your example above can be done more cleanly.

Hi David,

Many thanks, that's actually quite a good one. (yes, works really well .thanks!) I've also decided I want the worst possible and the best possible way for this.

Definitely your suggestion helps in the latter.

For the former, I open the text file in Openoffice, and type the "newfirstline" and then go through a bunch of dialogs in order to save as text. A good 5 minutes is required for this, and it is not automatable.

No, I write the first line on paper, I scan and perform OCR, which is then echoed out with a subshell (as David) cat'ng out the file and overwriting. Good if tedious 20 minutes.

You mentioned sed ...

Hi joeBuffer, thanks for the contribution.

I'm also following two tracks ... 1) fooling >> into prepending rather than appending 2) fseeking to a memory location earler than the file itself and inserting the firt line toegther with \n in there (if kernel will let me).

???

OpenOffice has API and, I think, Python interface, maybe bindings.

Hi Sergei,

OK, I retract. I am being facetious, you know ... joking.

Nearly everything is automatable in computing, but sometimes the effort required outweighs the benefits, a fact the theory often misses out on.

cheers.