Guillaume Demeyère <[email protected]> writes: > Hi Michael,
Hi Guillaume, > 11:49:31.806104 tramp-find-inline-compress (2) # Couldn’t find an > inline transfer compress command > > However, I think that is because I did not answer your question > accurately enough. The command: > (shell-command-to-string "echo xyzzy | bzip2 | bzip2 -d") > returns > "xyzzy > ", with *one space* after the last y, and one newline. > Thsi space causes the following line to fail: > (string-match > (concat "^" (regexp-quote magic) "$") (buffer-string))) > > This space that we get in the output comes, I think, from the space > between "xyzzy" and the first pipe. If you remove it ( echo > xyzzy|bzip2...), it disappears from the output. The newline does not > disappear, though. > > I did my tests with the following command: > (with-temp-buffer > (tramp-call-local-coding-command "echo xyzzy | compress | compress - > d" nil t) > (string-match > (concat "^xyzzy\s*$") (buffer-string))) > > and I checked that the command was successful only if I added \s or > \s* before the $. > > So I tried, in your patch, to replace "$" by "\s*$".... and it still > does not work. I also tried "\s*\n$", to no avail. It only works if I > completely remove the $. Grrr, MS Windows :-( Under GNU/Linux it works. So I have modified the expression to "^xyzzy", which is sufficient for our purposes. I've pushed the change to the repositories (slightly modified), will appear with the next Tramp release in GNU ELPA, in a couple of weeks. For Emacs 26.3 I keep the change to suppress compression on MS Windows. There exist already a pretest, and I don't want to damage anything. There's always the option to use Tramp from GNU ELPA. > I tried writing the buffer-string to *Message* in the patch, and it > seems the buffer-string has a "^M" (control-M ?) between the space and > the newline. I'm not sure where it comes from (I don't have it in the > test outside the patch), nor how to match it in a regexp (except by > completely removing the $... ?). Perhaps you could enlighten me about > that. MS Windows uses CR/LF as end-of-line indicator. "^M" is the CR char. > One more thing : I made the test with a newly installed Windows from > one of my colleagues, and the command "compress" works without the > need to install anything. So I did not need to install bzip2. And so > it makes sense to propose compression out of the box in Windows Good to know. > Regards, > Guillaume Best regards, Michael. _______________________________________________ Tramp-devel mailing list [email protected] https://lists.gnu.org/mailman/listinfo/tramp-devel
