Re: Hanging at mime-encode on Win2K -to- Tru64

[Nelson E Ingersoll]

Kai Großjohann wrote:
> 
> It is good that you have now found a value that works.  You seem to
> think that there is a size limit of 127 somewhere.  Perhaps that is
> not the case, and it is just a matter of time.  If you would like to
> investigate further, you could try to tweak the time it waits after
> each chunk.  With a larger wait time between chunks, it should be
> possible to increase the chunksize somewhat.
> 
> Currently, it sleeps for 0.1 seconds between chunks.  You can find the
> sleep-for call in tramp.el, in the function tramp-send-string.  You
> can just go to that file, edit the 0.1 into something else (0.2,
> perhaps, or 0.5 or 1.0) and then hit C-M-x with point still inside the
> defun.  Then you can try again if a larger value of tramp-chunksize
> works.

   I will try this.  I had to again stop using tramp and revert to the
simpler ange-ftp because tramp was still hanging.  Specifically when
saving buffers.  Emacs would go away and hang for a long time.  I am
using Emacs on Windows, and one symptom is that the windows would not
refresh when focus was changed away and back.  However, finally Emacs
would come back and the file was written.  Interestingly it also hung
when I set tramp's default method to "ftp".  I KNOW it was using
ange-ftp because it said so in the message buffer.  Things that make you
say, "Hmmmmm".

  The file was written; however, a 1-2 minute wait to save a 30K file
seems somewhat excessive.  Tramp is already very slow.  Anything at all
which adds to it's run time is, IMHO, horrible.  And 1-2 minutes is
outright unacceptable.   Your suggestion is worthy of following.

> You should kill the *tramp/foo* and the *debug tramp/foo* buffers
> between attempts, but other than that no restart of any kind should be
> required -- just tweak the defun, tweak the tramp-chunksize value,
> kill the buffers, and try again.  That is the beauty of an interactive
> environment like Lisp -- very rapid tweak and try cycles :-)

   Excellent!  I like faster turn around of effort.

> 
> You can use M-: (setq tramp-chunksize 42) RET to change the value of
> tramp-chunksize.  Note that I really mean M-: and not M-x.

   Actually I'd been doing C-X C-E at the end of "(setq tramp-chunksize
127)" in the *scratch* buffer.  So, I learned something new!  And so
early in the day too.  Wow, I can go home and take a nap.  Thanks! <grin>

- Nelson ...