Re: sequencer beta release

[Ben Franksen <ben.franksen@online.de>]

Hi Matthias

On Dienstag, 21. September 2010, Matthias Clausen wrote:
> On 17.05.2010 13:15, Benjamin Franksen wrote:
> > Hello All
> >
> > I would like to announce seq-2.0.99, a beta release for 2.1.0.
> >
> > The warning I issued in a previous mail regarding the future of
> > this development branch is hereby happily *withdrawn*. By general
> > consensus (of all interested parties) it was decided that I take
> > over maintenance and further development of the sequencer. So, this
> > branch *is* the new trunk, and it *will* be maintained and
> > (hopefully) further improved.
> >
> > The new sequencer home is at
> >
> > 	http://www-csr.bessy.de/control/SoftDist/sequencer/
> >
> > The main differences to 2.0.98 are:
> >
> > * converted the old Subversion repository at APS to darcs and
> > re-based my changes on this repo
> >
> > [NOTE: This means you should discard any clone you made of the
> > repository and get a new copy (via darcs get), as it will not be
> > possible to transfer patches form one to the other without major
> > conflicts. Sorry for any inconvenience this might cause -- I
> > decided it is more important that we have a continuous history in
> > the repo.]
> >
> > * integrated the few changes that were made in this repo after
> > 2.0.12 * major rework of the documentation, convert docs sources to
> > reStructuredText, use `Sphinx`_ to generate new website from the
> > docs * some minor cleanups, but otherwise no changes were made to
> > the code
>
> does this include the modification we made for remote management and
> redundancy support?
>
> We'll be happy to test your version - if so.

No, none of the code from DESY is integrated yet. My first goal is and 
was to first get 2.1 released. Unfortunately this has been stalled. I 
discovered a very deep bug in the sequencer (search for "Race 
conditions in SNL programs" in tech-talk archive, if you want to know 
the details). I was and still am adamant that the first release I make 
of the sequencer must provide some solution to this problem. I have 
thought very long and hard about this. It turned out that this cannot 
be fixed in a fully automatic way. Thus some old programs may still 
work unreliably, but from release 2.1 onward the SNL programmer will at 
least have the possibility to fix them and to avoid the problem in new 
programs. I have some code that I think does what I want but it needs a 
bit more testing and is not yet documented. Unfortunately I could not 
work very much on the sequencer during the last 2 months, but I hope I 
will soon get to it again. There will be at least one more beta release 
as these changes are much more tricky to verify than what I did with 
the compiler.

As soon as 2.1 is released, I will contact the person who did the 
changes at DESY and start to integrate them.

Cheers
Ben