Re: [XBoard-devel] interested in testing for you

[h.g. muller]

Perhaps it is a good idea if we indicate some areas that need testing.
For one, we specifically need testing of the development version
(the master branch in the Savannah repository). The released stable
versions (v4.4.x branch) automatically undgo heavy testing, exactly
because they are already used, and a lot of people use them.
Even the yet unreleased commits in this branch are almost exclusively
risk-free small fixes to reported bugs; there are no new featurse in this
branch.

In the development version, there have been a lot of large-scale code
changes, and the risk that something that used to work before got
broken is very real. Possible areas of concern:

1) The adudication of games to draws because of 3-fold repetition
or 50-move rule. This code has changed a lot, and is now also applied
in human vs engine mode, making it extra important that it is flawless.
Even in ICS play it is now used (although of course there the ICS decides
the game result), to handle some draw offers by the user as draw claims,
for ICS that have special syntax for claiming draws (such as FICS).

It would be good if people tested this in all modes (human-engine,
engine-engine, human-FICS), by setting up positions that are close
to drawing (e.g. set up a position by pasting in a FEN with a high
50-move counter), and then set engines playing or play a few moves by
hand, to see if the XBoard adjudicates the game at the right moment
(in accordance with the Options->Adjudcations settings). Or logging in
twice on FICS, to set up a game against yourself, and play some moves
that would lead to a claimable draw, and then test if a user draw offer is
handled as offer (relayed to the opponent, who can decline it) or as a
claim which is directly granted by the ICS.

Another concern is the interface code for move entering, which has also
been extensively changed. Do click-click moves, drag-drop moves all work
the same way as they did before? How about pre-move? Of special interest
are pathological situations. What if the opponent (engine or ICS) moves
while you were entering your (pre-)move? What if he resigns the game
while you are doing this, or loses on time? Will the next game start
properly, or will there be unresolved remnants of the interrupted move?

Yet another area is the saving of the settings. For XBoard this is entirely 
new.
Are there any undesired effects of saving the options in a ~/.xboardrc file,
like we do now?

In engine play, does the option to show the position at the end of the PV
by rightclicking the board or engine-output window work correctly? Can
you step through the PV by moving the mouse, and are all moves displayed
correctly?

Another new thing is the variation support. Can you properly walk the variation
tree of a loaded PGN that contained variations? Can you add variations to
a game in Edit Game mode? Are there any surprises or undesired side effects
of the variation support when you are analyzing a game with an engine?

For the ICS-mode improvements there is of course the Seek Graph, which
is entirely new. Are the seek ads properly displayed, and when you click them,
will you always play the person you selected? Does the dual-board option
work correctly when you play bughouse?

In general:
The new features of the development versions are described on
http://hgm.nubati.net/news.html. They all deserve some testing.
Not only if they work, but also if they interfere with existing functionality.