Re: [eboard-devel] Issues with EBoard

[Felipe B. <be...@se...>]

On Sun, 10 Jun 2001, Gian-Carlo Pascutto wrote:

Hello, thanks for the report.
I'll check everything in more detail,so this is just my first impressions
on everything.

> - It dumps prompts at such a rate any tells scroll out of
> the window in seconds

Are you using a non-standard prompt ?

> - Using the game list and observing doesn't work. I get
> 'no such game'. This _always_ happens. I do get a notification
> when the game ends.

I believe you are refreshing the game list often, right ? For it won't
refresh automatically.

> - I always expect the icon to go spinning and announce
> 'you have mail'

mail ma...@ti... -s XBoard Icon Patch

Hi Tim,

I've just finished this patch, xboard's icon will now say 'You've got X'
to /dev/dsp every now and then. It will take --x-files and --x-men as
arguments too.

Sincerely yours,
Felipe

P.S.: xboard's backend.c can be used to scare little children.
.

> - Dragging a piece produces very noticable flicker even
> with an XFree accelerated video card (Matrox G450)

When I first implemented dragging, I barely believed it would work at a
decent speed at all. I'm sorry to say, but on my US$ 20,00 PCI Diamond
Stealth 64 Trio64V (S3/865) with 2MB RAM it doesn't flicker at all.
Some friends of mine spent over US$ 500,00 buying nvidia
accelerated-3D-extra-ultra-power-64MB-video-cards just to see XFree86 fail
blatantly to run with them, or had to wait months for the drivers.
Buy a video card designed specifically to play Quake and Quake is all
you'll play on it.

But I'll try to reduce flickering somehow. Acceleration, eh ?

> - If I drop a piece after dragging, the piece flips
> back to the home square and eboard does the (animated) move
> itself. (extremely annoying, luckily does not happen during
> game play but does happen during analysis)

To be fixed.

> - I was examining a game and type 'follow'. Now I cannot
> get the window of the examined game to disappear.

One more to the my-stl-list-of-games-is-a-mess.

> - You cannot type into the detached console window.
> (this is annoying because I have to use the detached
> console to talk and play at the same time. But if I
> am not playing I pop up the console window so I can
> have more talk on screen. If I want to type something,
> I must switch windows each time)

Can be worked around. My particular experience is that when I want to
talk, I use the F-keys to switch to the console (F6 goes straight to the
console pane).

> - When you castle it highlights way too much squares.

It highlites all changed squares since last move. (Examining a game and
pressing the 'forward all' button can lead to a much more annoying result)

> - It does not allow you to toggle logging a session from within
> the interface.

To be fixed soon.

> - After I finished a game the seek info appeared in my
> console even though I had set eboard not to do this.

very weird, it isn't happening to me, may fall in the 'are you using a
non-standard prompt ?' category. If I recall, there's a way to place a
timestamp on the prompt. Are you using that ? It may really screw eboard
parsing...

> - When I do 'journal' some lines get colored and others
> do not with no appearant system in them. Similar things
> happen with other commands.

Must check.

> - At a certain moment I was playing a game, and because
> follow was still active another game suddenly popped up.
> I typed 'follow' 'unobserve' 'refresh' and eboard responed
> by crashing.

You should have no reason to ever type 'refresh' on eboard.

> - eboard cannot be compiled with GCC 3, from what look to be
> C++ standard violations

I'd really like to see an statistic of how much of the current C++ code
around is 'standard-compliant'. Languages bred in comittees tend to be
awful to use and die soon. K&R did an excellent job with C so ANSI
couldn't screw too much with it. Stroustrup created the whole skeleton for
OO and code reuse, but gave no code to be reused (something Sun worked
around much better with Java and its API). STL is a hack. A dirty one. The
std:: namespace thing is so bad that every compiler goes away from it in a
different style. What I call 'C++' is an ad-hoc standard I learned over
the years. Should GCC be as moronic as it seems it is, it's more probable
that I convert eboard to Eiffel or Objective C -- damn, even Perl -- than
making it compliant with the C++-standard-of-the-month. (I don't care what
a document says, the effective 'standard' is what the compiler compiles or
not - maybe if GNU came with GCC 3.0 10 years ago it would be standard by
now - but ask me again about it 5 years from now)

I'm sure GCC 3 _will have_ a command-line switch to accept "bad code".
Linus currently recommends egcs 2.91.66 for kernel compiling instead of
the newer ones. I wouldn't install a distro that came with a too-new
compiler on my box. I'm sure all 3 GNU/Hurd users will be happy to run the
latest version of GCC, but we Linux people can just get the best instead
of the latest. For people with fresh SuSE/Mandrake/RedHat/Conectiva
distros, there is always the option of a binary package. And for debian,
well, after I saw the results of

apt-get moo

I believe anything.

> - The piece antialiasing is pretty bad of quality. (at some
> sizes it does not do justice to the pretty piece bitmaps)

I agree, anyone have directions at a good (GPL'd) antialiasing code ?

> - There are bugs in the dragging-display algorithm. While
> dragging a piece, when the piece has left the home square
> about half, you can see the original piece being cropped
> out by a rectangle. This is ugly.

My only reply now is: better than not having dragging at all. But will
be fixed.

> - I wanted to move my king and changed my mind. To 'cancel'
> the move I dropped the king outside the board. It disappeared.
> (came back after making another move)

It's 10 am. Do you know where your king is right now ?

> - Even when premoving every move you still lose about 3 seconds
> for every 10 moves.

You got timeseal, right ?

> Except for that, it works fine.
>

No, there are many other bugs.

.........................................................................
Felipe Paulo Guazzi Bergo - Free Software Developer (be...@se...)
Personal Info and GPG Public Key: http://www.advogato.org/person/khazad
Campinas - SP - Brazil - Earth

* "God is dead" - Nietzche. "Nietzche is dead" - God.