Experimental webserver

[Derakon]

Lag isn't something I can solve with software - the server is located in Europe, and while I can play without noticeable lag here I have heard this a lot from people further away.

If you were really ambitious, you could split the server into a globally-distributed set of game servers sharing a common backend. Players would get the closest server, and spectators would get the backend (and concomitant lag). Put it up on AppEngine or something and you should be able to scale automatically to cope with high load, too!

I'm sure that's a perfectly reasonable project to work on in your spare time for no pay. And not at all overengineering.

[Gwarl]

Chat is in! Just about. You need to be logged in to see it. It's pretty rudimentary right now - there's one room and no list of users etc, but it's actually a frontend to an irc server so we can have all that. There are also mysterious server crashes for reasons I can't discern, so be careful with that and try not to break anything.

I've made some progress getting Sil to fit the mold and it should be available to play soon.

I will also say that while individual tasks are always easier than I would expect fitting it all together is more difficult than I nonetheless continue to expect. Easy access to char dumps and recorded games are next on the agenda.

How to tell if you've crashed the server: if you type something into the chat, and don't see your own message displayed back to you, please stop doing that and reload the page. The server will reboot itself on a crash, but it's still not a very nice thing to do.

Consider the current implementation of chat to be 'it might work, if you're lucky and the tides are right'

[Pete Mack]

What happened to the Borg? I haven't been able to watch it in recent builds.

[Pete Mack]

Grrr. Died to a dark elven sorcerer, entirely because it was purple instead of light red. These colors are not user friendly.

[HugoVirtuoso]

The webserver colors for PosChengband are a big off also

Edit - Sorry, I meant to say "The webserver colors for PosChengband are a bit off also"...NOT "a big off"....

[MITZE]

Bug?

Just created a profile and started a Angband character; kobold warrior, and apparently I know about amulets of adornment, rings of protection, and rings of reckless attacks from birth. Is this something that should be happening?

[Estie]

Just created a profile and started a Angband character; kobold warrior, and apparently I know about amulets of adornment, rings of protection, and rings of reckless attacks from birth. Is this something that should be happening?

No thats all right. Amulets of adornment do nothing, and the other 2 give boni to AC, to hit, to damage which are known because basic items have it. You just happened to find the known mods first.

There was a time when you didnt know that your starting dagger was (+0, +0) and instead it was (??). That was weird; this way is better.

[Pete Mack]

Yeah, it'd be possible that (+0,+0) would not trigger the ?? flag and initially just show as a dagger (1d4). But the melée and armor bonuses are just too easy to figure out to make it worthwhile. Literally the first monster you hit (or hits you.) Brands are essentially as easy of course, but it's no effort to make them work as normal.

[Gwarl]

Yes that's as expected - the version of Vanilla angband installed on the server is a development build from about a week ago, and the ID game has changed since 4.0.5

[Gwarl]

Okay I think I've fixed the issues with the sockets timing out - feel free to tab away and come back. Chat should be working too but bear in mind it's the crudest implementation imaginable - the irc backend has been done away with for now, but it will come back.

[Gwarl]

I've been looking at colours and I need some help.

I can set the colours on the terminal emulator by passing in an array of hex strings when initialising it. There are two default colour sets, calling themselves tangoColors and xtermColors. The problem is, what angband thinks is light red, the ordering of colours in the terminal emulator clearly intends to be magenta.

To be specific, the colour R in angband terms is seen as colour 13 in the colors array (counting from black=0). If R -> 13 is meaningful to angband in any way and I could determine the mapping of letters to ints I could explicitly set all 28 colours. If not I'll have to look into it more deeply.

[Gwarl]

Okay, z-color.h was what I was looking for. I'll get to work.

Edit: I think the odd colours may actually be angband's fault. I found this in main-gcu.c line 982:

		colortable[COLOUR_L_RED]    = (COLOR_PAIR(PAIR_MAGENTA) | A_BRIGHT);

Okay yeah angband thinks we have 8 colors available so it's trying to work with them. If I can tell the terminal emulator to pretend it's a 256 color terminal not an 8 color terminal we should be okay.

Triple edit: No I'm hopelessly confused now. We're defining 8 color pairs, and mapping each of our 28 colours to a colour paired with black. Why?

Final edit: I think we should have decent color support now

Encore edit: Bear with me

Okay I think colors are fixed. shift-running in PCB is still impossible for now but I'm looking into it.

[MITZE]

Webserver works!

Just had a nice afternoon playing PosCheng while chatting with Gwarl--the webserver works!

[debo]

Did the server move away from angband.live, or is nodejs just dead or something? I'm getting 502s from nginx.

[Gwarl]

Did the server move away from angband.live, or is nodejs just dead or something? I'm getting 502s from nginx.

nodejs server died. Sorry. Back now.