Re: [Tlf-devel] bugs in tlf-1.2.0_pre2

[Thomas Beierlein]

Hi Martin, hi Fred,

Am Mon, 10 Sep 2012 11:44:41 +0200
schrieb Martin Kratoska <address@hidden>:

> Hi Fred,
> 
> although I cannot find it (now) in the source code, I suspect the
> intial work of Rein, PA0R, which included some "hardwired" contests.

You are quite right. Nearly all contest logic in TLF is hardwired. That
goes for counting points and multipliers, what is to be displayed in
the summary, the calculation of claimed points, the format of the
log and the cabrillo file and much much more. 

While the code for that contests do mostly work the problems are in the
area of self defined contests. Rein did some work to allow giving more
or less points to different class of QSO's and also a start of a
multiplier logic (WYSIWYG and others). But that always solved problems
only for special contests. So by time there were more and more special
rules, which do not fit together  for modern contests. As Martin said
it must be fixed soon.

> If tlf should have any future, it must be cleaned from the remnants
> of hardwired contests, also it must be freely configurable.

Right. But it is a lot of work and time is sparse.

The actual work on the 1.2 series is a start in that direction. It will
have two main new features:

- record actual frequencies from rig as requested by more and more
  contests (that part is done and can be tested in 1.2.0_pre1 and pre2)
  and
- a complete new routine to write the cabrillo logs. 

  (First as you both pointed out, the actual implementation of
  write_cabrillo is broken and works only for some special contests. 
  Second we have some new contests which are not covered by the routine
  at all.)

  The new implementation will be controlled by a textual description of
  the cabrillo line which points out what to put in the line, where to
  put it and how much characters the field should be have (on github
  you can have a first look at that description syntax in
  share/cabrillo.fmt).
  That format description can be extended by each user of tlf so we
  can easily adapt new contests not covered in tlf's code.

I am just on the way to complete the second part. I hope to finish it
in the next two weeks. That should fix the wrong cabrillo formats and
is a first step to make the whole tlf completely configurable.


To make the whole tlf contest logic free configurable there are some
more steps to go:

- specify rules to give points to qsos and to count multis (What are
  points given for, how to handle unequal points for different bands,
  modes, for low power stations and so on. What counts as a multi,
  which possibilities are needed to have uncommon multi rules....)
- specify what to display on the summary panel and how to calculate the
  total score
- find an easy way to formulate theses rules in a consistent text file.


> The mess can be tracked down in the rule files. Although you define
> your own contest, your rule file still needs to begin with a
> CONTEST=**** directive which has ***PROBABLY*** another meaning than
> just a contest name. 

No, in the rules file that information is quite useless. It has only to
go to the logcfg.dat. It is just in the rule file to make sure it is
the right file.

> If you put there CONTEST=cqww you will get a
> hardwired (predefined) DXCC and CQZ mults. In other hand, you don't
> need to define any mult, the hardwired items are still there.
> 
> My sincere apologies if my conclusions are totally false, actually I 
> have a mess in my files since tlf 0.9.3 and my programming (un)skills 
> preventing me from more thorough research of tlf internals. I hope
> that some day a tlf expert comes with a deep and thorough explanation
> of the tlf mechanics.
> 
Yeah. tlf's internal working is quite a hardcoded mess. It took me
nearly a year to understand its working (at least I hope so). The
whole logic is scattered around a lot of places which makes it
difficult to change a single piece without doing damage on other things.

In the last months I did a lot of cleanup to uncouple some of that code
and make it more maintainable. but it is far from done.

73, de Tom DL1JBE

> 73
> Martin, OK1RR
> 
> Dne 10.9.2012 11:00, FS napsal(a):
> > The interesting thing is that, in some contests like CQWW this
> > works, in some not. If it works the number is printed with 4 digits.
> >
> > 73 Fred
> >
> > On 09/10/2012 01:19 AM, Martin Kratoska wrote:
> >> The September CWopen by CWops discovered following bugs in
> >> tlf-1.2.0_pre2:
> >
> >> Another bug - if defining the part of exchange containing the
> >> serial number (with #), the serial is actually not put into log,
> >> the # character appears instead.
> >
> > _______________________________________________
> > Tlf-devel mailing list
> > address@hidden
> > https://lists.nongnu.org/mailman/listinfo/tlf-devel
> >
> 
> 
> _______________________________________________
> Tlf-devel mailing list
> address@hidden
> https://lists.nongnu.org/mailman/listinfo/tlf-devel



-- 
"Do what is needful!"
Ursula LeGuin: Earthsea
--