Angband and Ubuntu

[Uqbar]

Hi all.
The current situation with Ubuntu is that "they" only support v3.3.2 which is 3+ years old, even on bleeding edge builds like Vivid Vervet (aka 15.04).
It's a shame that has been already reported elsewhere.
In the meanwhile I have somehow managed to compile angband-3.5.0 under my ubuntu 14.04 (LTS). No GTK, no SDL, just nCurses (when I ssh) and X11 (when I'm local).

I have a couple of issues:

1. Exact support file placement. I would like to have my build store those files in /etc/angband just like the official Ubuntu build. I build

2. System wide file permissions. It looks like the installer part of the makefile won't take them into any account, so I have to manually fix it.

3. It seems the v3.3.2 save files are incompatible with v3.5.0's, so "upgrading" won't allowe players to restore those games.

For no.1 it's more about source configuration.
At the moment I have this configuration command:

./configure --enable-x11 --enable-curses --prefix=/usr --sysconfdir=/etc --with-private-dirs

I am almost there, but not everything fits in the proper directory.

For no.2 I manually run a script to fix permissions.
My system's root user has a restrictive umask set in its .bashrc file, as per best practices. It'd be much better to have the installation procedure to explicitly make directories like "755" and files like "644" instead of relying on some "default" (?) root umask.

For no.3 I have no idea whether there can be a solution.

Any hint?
TIA.

[Nick]

I can confirm that there isn't a solution to no 3. Savefile compatibility is not broken between versions unless there's no way around it, in which case characters from the two are not really comparable; you're better finishing your old character in the old version and then upgrading for subsequent characters.

Ubuntu uses the Debian package. The gap in the versions is due to the difficulty of getting a correct Debian package accepted by Debian, and the time spent waiting for the next Debian release to actually happen.

Your other questions are better answered by people other than me

[Uqbar]

What I'd need for v3.5.0 compilation is "just" which configuration options are needed in order to get it "Debian/Ubuntu" ready.
I don't mind/want to submit that build to anyone but my servers.
I have a number of game saves that I like to use to train. It's not just "finishing" them. It's more about keeping.

[Magnate]

Apologies - I am the Debian maintainer for Angband and RL has prevented me packaging 3.4.x or 3.5.0

I am also the release manager for 3.5.1 and will package it as soon as it's released (three bugs left, one of which is unreproducible).

[Malak Darkhunter]

So I just got to ask, how long before next release? sounds like you guys are awfully close.

another question I am really curious about, if I run angband with x11-and ncurses it looks a lot different than say from another variant, take Sangband for example..it's not nearly as nice looking in the terminal as Angband is, have you guys made changes to the look and feel of x11 and ncurses in Angband or are you just using newer libraries?

[Nick]

So I just got to ask, how long before next release? sounds like you guys are awfully close.

3.5.1 will be out before too long - it's 3.5.0 with the bugs fixed.

Meanwhile we are still going on the major restructure, which will eventually be released as 4.0. It also will have (almost) no gameplay changes; we will release an alpha/beta/work-in-progress/whatever at some point to help get the bugs ironed out.

[Magnate]

3.5.1 will be out before too long - it's 3.5.0 with the bugs fixed.

Meanwhile we are still going on the major restructure, which will eventually be released as 4.0. It also will have (almost) no gameplay changes; we will release an alpha/beta/work-in-progress/whatever at some point to help get the bugs ironed out.

As RM for 3.5.1 I am *intending* to release it over the xmas break, before I go back to work on 5th Jan.

No guarantees mind ....

[AnonymousHero]

Apologies - I am the Debian maintainer for Angband and RL has prevented me packaging 3.4.x or 3.5.0

I am also the release manager for 3.5.1 and will package it as soon as it's released (three bugs left, one of which is unreproducible).

Why are we waiting for three bugs in particular? Aren't there enough bugfixes already in master to release 3.5.1 and then follow up with a 3.5.2 at some point? If there release process is too arduous for this to be practical, then I submit that the release process needs to be revisited and optimized.

(I recognize that things move slower in Debian-world, but that's no reason to delay general Angband releases.)

EDIT: Oh, and this is not meant as snark or anything. I recognize that doing (releases of) F/OSS is thankless work, but that problem can be reduced by making release management more push-button.

[Nick]

Why are we waiting for three bugs in particular? Aren't there enough bugfixes already in master to release 3.5.1 and then follow up with a 3.5.2 at some point?

Quick answer is we want to focus on 4.0 - even doing 3.5.1 was not even certain, and we're all very grateful to Magnate for handling it. In particular, I've put pretty much zero effort into 3.5 post 3.5.0, and with the restructure branch now more than 1000 commits past 3.5.0 I'd be a bit of a liability on 3.5 code.

[AnonymousHero]

Quick answer is we want to focus on 4.0 - even doing 3.5.1 was not even certain, and we're all very grateful to Magnate for handling it. In particular, I've put pretty much zero effort into 3.5 post 3.5.0, and with the restructure branch now more than 1000 commits past 3.5.0 I'd be a bit of a liability on 3.5 code.

But aren't there enough commits/fixes regardless of anything happening in a side branch? Or is Magnate being set up for cherry-picking a huge amount of commits or something? (If the latter is the case, then I think there's an overall issue with the way branching is being handled.)

[takkaria]

Why are we waiting for three bugs in particular? Aren't there enough bugfixes already in master to release 3.5.1 and then follow up with a 3.5.2 at some point? If there release process is too arduous for this to be practical, then I submit that the release process needs to be revisited and optimized.

But aren't there enough commits/fixes regardless of anything happening in a side branch? Or is Magnate being set up for cherry-picking a huge amount of commits or something? (If the latter is the case, then I think there's an overall issue with the way branching is being handled.)

I don't think there's anything wrong with our release process. No-one is cherry-picking huge numbers of commits and we defined 'enough fixes' at our last devteam meeting as 'when bugs A, B, C ... X, Y & Z were fixed'. They're not all fixed yet, hence no release. FWIW there are 43 other commits on master ready to go, amassed over the last year.

Hang in there, it's almost out

[AnonymousHero]

Hang in there, it's almost out

I usually run from 'master', I was just curious .

[Magnate]

I usually run from 'master'

I think that puts you in a fairly small group of players.

FWIW I think the release process is fine, but it is definitely not push-button - and never will be. To make a half decent release you need to check or update various things manually (changelog, thanks, FAQs) and do invisible but important things like git tagging etc. We got it down to this, which is ok, but definitely something I'd rather not do twice. 3.5.1 will be the last release "before the restructure" - ah good, I'd been wondering what to call it ...

[takkaria]

I usually run from 'master', I was just curious .

I said master - I meant 3.5-release. Oops.