2010-11-10



I have written nothing for one week.  Not  because  I  didn't  do
anything,  in fact I did a lot, but rather because it hadn't been
special or I hadn't been in the mood to write.

My crawling though nmh's code base goes well.  I really  can  say
that  my  current  knowledge of nmh is a magnitude better than it
had been just a month ago.

I wrote several mails to the nmh-workers mailing list,  some  in-
cluded  small  patches.  To my delight two are already applied by
Peter Maydell.

Although I don't mind too much, it would be nice to read my  name
in  the  ChangeLog  or in the commit messages for stuff I brought
up.  I mean, although these improvements were small, I am  pretty
proud that I could discover or initiate them.  These are my first
real contributions to nmh.  Some work I do on  Free  Software  is
for  my own needs, some is for the community only (like improving
the man page of slocal(8) which I don't use myself).  Much of the
motivation  in  Free  Software comes from acknowledging any work.
Acknowledgement fuels the Free Software community.

I asked Peter Maydell, who's a nice guy btw, about tasks I  might
go  for  and  he told me some.  Someone who has the overview is a
good advisor for planing ones steps, when you have a very limited
view.

Understanding (most parts) of send(1) and post(8) was a mayor im-
provement for understanding nmh.

Sendfiles(1) and viamail(8) were nice discoveries.

The more I think about the implementation of  MIME  in  nmh,  the
more do I believe Jon's attachment approach (via headers that get
processed in send(1)) is the way to go.  It at  least  can  cover
comp(1)  and forw(1).  For repl(1) a ``give me all (or the first)
text parts, decoded please'' facility needs to be set up.  That's
if  on  the sending side.  For displaying, show(1)/mhshow(1) need
probably be redesigned anew.



Besides nmh, I had a nice meeting with  awk(1).   I  came  across
this trim function:

     awk '{gsub(/^ +| +$/, "")}1'

It removes spaces (should probably tabs too) from  the  beginning
and  end  of the line.  But what does the `1' at the end?  I know
awk pretty well but still had been  confused.   In  fact  it's  a
language  abuse.   It's easier to understand it if the program is
formated nicely:

     awk '
     {
             gsub(/^ +| +$/, "")
     }

     1
     '

The program consists of two blocks.  The first has  no  condition
and  thus  runs  on every line, trimming the leading and trailing
spaces.  The second block, and that's the point, consists of only
a  condition  (`1') but no actions.  `1' simply gets evaluated as
true and for missing actions  the  default  action  `{print}'  is
used.

I can't decide if I find this abuse of awk wonderful or terrible.
;-)

Anyway, it would have been clearer, and thus better, to write:

     awk '{gsub(/^ +| +$/, "");print}'

or take sed(1):

     sed 's,^ *,,;s, *$,,'



Sometimes a goto is the better choice.  (I really do believe so.)
Like  we  have these unstructurated break and continue statements
and early return, we also have the goto.  All these are  valuable
at  the  right  time.   For  some (e.g. the early return) this is
often the case, for others (e.g. the goto) it's seldom, but there
are cases.  Goto gives great power to the programmer, it's on him
to use it wisely.  But not at all should goto be a no-go.



http://marmaro.de/lue/        markus schnalke <meillo@marmaro.de>