None of the old Angband variants that I used to play so enthusiastically
on my A5000 seem to function on an ARMX6 even under Aemulor, despite the
fact that they are basically ASCII text displayed in a desktop window.
For example, Steamband
http://angband.oook.cz/steamband/steamband.html
will run the character generator when you start a new game, but exits
with a silent 'abort on data transfer' error as soon as it tries to
start the actual game (I suspect as soon as it tries to create the save
file, since nothing is saved).
I'm not all that interested in getting the latest Angband developments
ported to RISC OS - many of the old variants have long since ceased development anyway - but it would be nice to be able to get back some of
the plethora of ports that used to be available, all of them so far as I remember based off a single 'Angband compiling kit' that provides a
standard WIMP front end. https://rec.games.roguelike.angband.narkive.com/j3kwr1kh/announce-risc-os-ports
Looking at the Angband forums, where all the prior Usenet activity now
seems to reside, I see postings like "There comes a time in all
development projects where legacy support must be cut, and you must make
your users move on.[...] I think it's reasonably safe to say that
support for RISC is rather... unnecessary." :-( http://angband.oook.cz/forum/showthread.php?t=2484
Harriet Bazley wrote in comp.sys.acorn.games:
None of the old Angband variants that I used to play so
enthusiastically on my A5000 seem to function on an ARMX6 even under
Aemulor, despite the fact that they are basically ASCII text displayed
in a desktop window. ...
I'm not all that interested in getting the latest Angband developments
ported to RISC OS - many of the old variants have long since ceased
development anyway - but it would be nice to be able to get back some
of the plethora of ports that used to be available, all of them so far
as I remember based off a single 'Angband compiling kit' that provides
a standard WIMP front end.
https://rec.games.roguelike.angband.narkive.com/j3kwr1kh/announce-risc- os-ports
Looking at the Angband forums, where all the prior Usenet activity now
seems to reside, I see postings like "There comes a time in all
development projects where legacy support must be cut, and you must
make your users move on.[...] I think it's reasonably safe to say that
support for RISC is rather... unnecessary." :-(
http://angband.oook.cz/forum/showthread.php?t=2484
I've managed to trace Musus Umbra's original Angband porting package http://www.zen22994.zen.co.uk/musus/angband/compiling.htm but that was
the generation before Antony Sidwell's "32-bit" ports - which are the
ones that don't seem to work on the ARMX6's Cortex-A9 processor - so
probably not a good basis for recompilation. :-(
Antony Sidwell's site http://ajps.mine.au and its Angband-Kit no longer
seem to exist.
Am Tue, 03 Aug 2021 19:10:12 +0100 schrieb Harriet Bazley:
Harriet Bazley wrote in comp.sys.acorn.games:
None of the old Angband variants that I used to play so
enthusiastically on my A5000 seem to function on an ARMX6 even under
Aemulor, despite the fact that they are basically ASCII text displayed
in a desktop window. ...
As long as an Emulator runs on the new machine I think it should be
possible to use the old versions directly.
I've managed to trace Musus Umbra's original Angband porting package http://www.zen22994.zen.co.uk/musus/angband/compiling.htm but that was
the generation before Antony Sidwell's "32-bit" ports - which are the
ones that don't seem to work on the ARMX6's Cortex-A9 processor - so probably not a good basis for recompilation. :-(
Antony Sidwell's site http://ajps.mine.au and its Angband-Kit no longer seem to exist.
That's really a shame if so many sites are gone and nobody ever made a
copy wich allows to reconstruct them.
I only had the version that was availabe from StuttgartFTP - it worked perfectly on A5000 and other 3.1 machines. Never ever tried to use
Angband on a new machine.
Oooh, at a quick test Musus Umbra's 2001 port of vanilla Angband 2.9.3
*does* work under Aemulor, even if the 2004/2008 variants don't: http://www.zen22994.zen.co.uk/musus/angband/a293.zip
It's quite impressive how much muscle memory of the keystrokes I still
manage to retain!
And I've got my old copy of Psiband (1999) working... so maybe I need to
do some experimentation to see exactly which variants work and which
ones don't....
On 7 Aug 2021 as I do recall,
Harriet Bazley wrote:
[snip]
Oooh, at a quick test Musus Umbra's 2001 port of vanilla Angband 2.9.3Apparently I was incredibly lucky in picking those two at random. :-(
*does* work under Aemulor, even if the 2004/2008 variants don't:
http://www.zen22994.zen.co.uk/musus/angband/a293.zip
It's quite impressive how much muscle memory of the keystrokes I still
manage to retain!
And I've got my old copy of Psiband (1999) working... so maybe I need
to do some experimentation to see exactly which variants work and which
ones don't....
Quick summary: Using Aemulor Ultimate, I can run Frog-knows (the
original RISC OS port of Angband), Umoria (the predecessor of Angband), Cthangband, Discband, Psiband, Kamband and Sangband. The very obvious
thing all these have in common is that they were all ported by Musus
Umbra on or before the year 2000. However Kangband (ported 1999) won't
run, and Angband 2.9.3 (ported 2001) will.
Vanilla Angband 2.9.3 works. Vanilla Angband 3.0.9 doesn't.
I also have in my possession RISC OS ports of Gumband, CaTHband, Eyband, Kangband, sCthband, NPPAngband, Oangband, Quickband, Steamband and
Zangband, all of which fail in various ways and many of which fail in
the same way (either crashing when they attempt to read/write savefiles,
or when they attempt to parse the plaintext files that define the game).
I don't have the source for any version of Angband, with or without the #ifdef RISCOS lines that were scheduled for removal in 2009....
Angband 3.0.9 (2007) complains about errors in 'terrain.txt' (the line
in question looks fine to me, or at least consistent with the format of
all its neighbours) and crashes.
Gumband (2008) complains about an error in the f_info.raw file, which is
the file generated from the terrain.txt file, and crashes.
CaTHband (2001) crashes with an abort on data transfer before it manages
to display any window at all.
Angband 2.4 Frog-knows (the original RISC OS port from 1993) works - continuing the pattern of the older ports being more compatible!
Umoria (1992) works, and apparently I still have a level 26 character
alive in it - unfortunately I can't remember what to look out for and
how to play. :-(
Eyangband (2003) crashes with the error "Cannot parse 'limits.raw' file"
- there's clearly a pattern here, but none of the corresponding data
files being used to generate the raw files on startup have been altered
since I last played the game, and they are all human-readable plain text files. So why should a change of processor affect the ability to use
them?
To my astonishment, Cthangband (1999), one of the most non-standard
variants, appears to work - I managed to play through several levels
before I was killed. Again, I can only guess that it's because it's one
of the oldest ports.
But then Kangband (also 1999), also very non-standard, fails to run with
the error "MususIfType not found", despite the fact that the MususUtils module is listed as currently running under Aemulor.
SCthband (2005) allows you to enter the character creation screens, but
then exits with a silent (visible only in Reporter) abort on data
transfer error at the point where it tries to write the savefile and
start the game proper.
Discband (2000) works, although it's simply a copy of Zangband 2.4.0
with the uniques, potions and scrolls renamed with a Pratchettian
flavour - the application inside the Discband directory is even called !Zangband.
Kamband the wildly unbalanced (1999) works, amazingly - I ran a quick
warrior character and was killed by my own pets firing arrows at our
enemies, which I seem to remember is a very common hazard. The AI is somewhat lacking!
NPPAngband (2006) gives the same mysterious terrain error; these
variants are simply crashing on the first line of terrain encountered in
the file.
Oangband (2005) crashes at the end of the character creation process
with "Unknown signal number 8", which is a new one!
Quickband (2005) crashes on the terrain file.
Sangband (1999) works.
Steamband (2008) either crashes with a 'broken save file' error, or
exits silently with an 'abort on data transfer' error when it attempts
to create a new savefile.
My copy of the Unangband application seems to have vanished from its directory at some point, leaving behind only an orphaned savefile!
And Zangband (2003) crashes with the error 'Cannot parse misc.raw file', which is presumably a variant on the various other mysterious failures
to initialise data files from text files. It's probably a very simple issue, like a problem with line endings or Unix directory separators, or
an attempt to identify the current version of RISC OS that assumes
anything above 4.x is an error. :-(
(Judging by the datestamps, all the variants fail on the first data file
they attempt; they just attempt to start with different files).
On 7 Aug 2021 as I do recall,
Harriet Bazley wrote:
[snip]
Oooh, at a quick test Musus Umbra's 2001 port of vanilla Angband 2.9.3Apparently I was incredibly lucky in picking those two at random. :-(
*does* work under Aemulor, even if the 2004/2008 variants don't:
http://www.zen22994.zen.co.uk/musus/angband/a293.zip
It's quite impressive how much muscle memory of the keystrokes I still
manage to retain!
And I've got my old copy of Psiband (1999) working... so maybe I need
to do some experimentation to see exactly which variants work and which
ones don't....
Quick summary: Using Aemulor Ultimate, I can run Frog-knows (the
original RISC OS port of Angband), Umoria (the predecessor of Angband), Cthangband, Discband, Psiband, Kamband and Sangband. The very obvious
thing all these have in common is that they were all ported by Musus
Umbra on or before the year 2000. However Kangband (ported 1999) won't
run, and Angband 2.9.3 (ported 2001) will.
Vanilla Angband 2.9.3 works. Vanilla Angband 3.0.9 doesn't.
I also have in my possession RISC OS ports of Gumband, CaTHband, Eyband, Kangband, sCthband, NPPAngband, Oangband, Quickband, Steamband and
Zangband, all of which fail in various ways and many of which fail in
the same way (either crashing when they attempt to read/write savefiles,
or when they attempt to parse the plaintext files that define the game).
I don't have the source for any version of Angband, with or without the #ifdef RISCOS lines that were scheduled for removal in 2009....
Am Sat, 07 Aug 2021 19:36:46 +0100 schrieb Harriet Bazley:
Quick summary: Using Aemulor Ultimate, I can run Frog-knows (the
original RISC OS port of Angband), Umoria (the predecessor of Angband), Cthangband, Discband, Psiband, Kamband and Sangband. The very obvious thing all these have in common is that they were all ported by Musus
Umbra on or before the year 2000. However Kangband (ported 1999) won't run, and Angband 2.9.3 (ported 2001) will.
Vanilla Angband 2.9.3 works. Vanilla Angband 3.0.9 doesn't.
I never get the point why there are so many different types of this
game ... Sangband, Kamband, Zangband and many more.
But eventually it could be a good idea to bring them back to the
internet, eventually at iconbar or another place where people will be
able to find and download them.
Some of the errors really seem to bes very simple - eventually sometimes
the files only need to be in existence but aren't there. Eventually they
have been there - but were lost in the ZIP files if they had 0 bytes
length.
Others are eventually not found because of typos in the source code.
I've tried to find some of the websites You mentioned - wich are all
gone. Many of them seems to be gone forever. But one of the main sites
tries to stand the test of times on
https://web.archive.org/web/20110927013626/http://rephial.org/
release/3.0.5
There are some other versions and times too. I used this one because it
seems to be one of the relevant Ports to RISC OS - and it comes with its sources - and all files found their way to archive and are downloadable.
At least there is a ReadMe File wich is signed by Anthony - hope this is
a sign of qualitiy, You cited an Anthony in Your posts, see above.
The Thangorodrim Site - wich seems to be the precursor site to
rephial.org is also available on
https://web.archive.org/web/20210622152216/http://www.thangorodrim.net/
There are many timepoints when archive tock a "shot". So eventually there
is enough info to get some hints.
Antony Sidwell's site http://ajps.mine.au and its Angband-Kit no
longer seem to exist.
That's really a shame if so many sites are gone and nobody ever made a
copy wich allows to reconstruct them.
It's possible the site still exists on the Wayback Machine, but that's
no longer searchable from Netsurf. The URLs still work once a site has
been located, but if I try to search I just get "The Wayback Machine
requires your browser to support JavaScript".
In article <e1824b5859.harriet@bazleyfamily.co.uk>,
Harriet Bazley <harriet@bazleyfamily.co.uk> wrote:
Antony Sidwell's site http://ajps.mine.au and its Angband-Kit no
longer seem to exist.
That's really a shame if so many sites are gone and nobody ever made
a copy wich allows to reconstruct them.
It's possible the site still exists on the Wayback Machine, but that's
no longer searchable from Netsurf. The URLs still work once a site has
been located, but if I try to search I just get "The Wayback Machine
requires your browser to support JavaScript".
The Wayback Machine search works on the Iris browser on RiscOS but sadly
the ajps site is reported as not archived.
On 8 Aug 2021 as I do recall,
Sebastian Barthel wrote:
Am Sat, 07 Aug 2021 19:36:46 +0100 schrieb Harriet Bazley:
[snip]Quick summary: Using Aemulor Ultimate, I can run Frog-knows (the
original RISC OS port of Angband), Umoria (the predecessor of
Angband), Cthangband, Discband, Psiband, Kamband and Sangband. The
very obvious thing all these have in common is that they were all
ported by Musus Umbra on or before the year 2000. However Kangband
(ported 1999) won't run, and Angband 2.9.3 (ported 2001) will.
Vanilla Angband 2.9.3 works. Vanilla Angband 3.0.9 doesn't.
I never get the point why there are so many different types of this
game ... Sangband, Kamband, Zangband and many more.
Mainly because lots of people wanted to create their own improved
version of the game ...
Quickband was an ingenious variant designed for those times when you
don't feel like investing several weeks/months in repetitive farming for artifacts in an attempt to complete 100 levels of Angband; it only
provides 12 dungeon levels, actively encourages the player to dive fast, implements a steeper progression curve, and does some major rebalancing
in order to make this compressed difficulty format work and eliminate
the large amounts of 'scumming' inherent in standard Angband play. (This
is one of the few variants where it would be nice to have a fresh RISC
OS port of the later versions, since it went on being tweaked for
balance until v2.06 of 2012, and we only had a port of v1.05 from 2008.)
But eventually it could be a good idea to bring them back to the
internet, eventually at iconbar or another place where people will be
able to find and download them.
It would be nice to have a repository of RISC OS Angband versions and variants in order to provide a central resource, and the Iconbar would
be a logical place, given that it already has an Angband section dating
back to Musus Umbra's day: https://www.iconbar.com/Angband_support_-_Introduction/news973.html
I'm happy to provide the various compiled versions I have, although none
of them appear to be 32-bit safe, or at least not ARMX6-safe. I suspect
that it would be hard to locate source code for most of them, but it
would be very nice to be able to recompile at least some of them so that
they work without Aemulor - or, indeed, work *with* Aemulor!
And one really ought to have the base game (Vanilla Angband) working for comparison, and it's currently one of the ones that doesn't.
Neither Antony nor Andrew Sidwell seem to have been active on the
Iconbar for a long time: https://www.iconbar.com/forums/profile.php?username=ajps https://www.iconbar.com/forums/profile.php?username=takkaria
I've tried to find some of the websites You mentioned - wich are all
gone. Many of them seems to be gone forever. But one of the main sites
tries to stand the test of times on
https://web.archive.org/web/20110927013626/http://rephial.org/
release/3.0.5
There are some other versions and times too. I used this one because it
seems to be one of the relevant Ports to RISC OS - and it comes with
its sources - and all files found their way to archive and are
downloadable.
At least there is a ReadMe File wich is signed by Anthony - hope this
is a sign of qualitiy, You cited an Anthony in Your posts, see above.
That version doesn't run for me - does it run for you? 3.0.3 doesn't
work for me either, but 3.0.0 does, and in fact without requiring
Aemulor... which incidentally proves that it's not the introduction of
Lua scripting for the 3.0.x branch that is causing the problem, nor the v3.0.0 comment "replaced the old Acorn (RISC OS) code with a newer
version"!
It looks as if version 3.0.9, which is the latest (non-working) one I
have, was the final release to include RISC OS; the release notes for version 3.2.0 include the statement "Remove RISC OS port".
The actual 'terrain.txt' file to which version 3.0.9 is objecting
appears to be identical to the one which works quite happily in version 3.0.0. :-(
Am Sun, 08 Aug 2021 03:06:09 +0100 schrieb Harriet Bazley:
I'm happy to provide the various compiled versions I have, although none
of them appear to be 32-bit safe, or at least not ARMX6-safe. I suspect that it would be hard to locate source code for most of them, but it
would be very nice to be able to recompile at least some of them so that they work without Aemulor - or, indeed, work *with* Aemulor!
Source is the primary problem. Secondary is to find a skilled person who
is willing to do such a "conversion". E.g. for Quickband and the last
version of Angband known to work on RISC OS.
If the sources can be found, it shouldn't be "too hard" for an
experienced programmer - as far as I understood, there is a RISCOS Layer
wich doesn't interact to closely with the game code.
But I will try some of these many "Xngbands" in the next few days/weeks.
But since I use !RPCEmu more often than RPi I think it will run out of
the box. We'll see.
Many thanks for the massive Links. Angband - seems to be a little
software universe in itself.
Am Sun, 08 Aug 2021 10:35:07 +0100 schrieb Alan Dawes:
In article <e1824b5859.harriet@bazleyfamily.co.uk>,
Harriet Bazley <harriet@bazleyfamily.co.uk> wrote:
Antony Sidwell's site http://ajps.mine.au and its Angband-Kit no
longer seem to exist.
The Wayback Machine search works on the Iris browser on RiscOS but sadly the ajps site is reported as not archived.
That's not true !
It is archived - but the name ends with .nu (not .au)
At
<https://web.archive.org/web/20090130011516/http://ajps.mine.nu/angband/ports/index.htm>
You can find all the versions that are made with Archimedes/RiscPC in
mind. So at best it runs with !Aemulor or !Archie or !RPCEmu.
The version I mentioned above and wich had been installed "here" is the topmost - the "Red Dragon" one. This one did fine - ages ago.
On 8 Aug 2021 as I do recall,
Sebastian Barthel wrote:
Am Sun, 08 Aug 2021 10:35:07 +0100 schrieb Alan Dawes:
In article <e1824b5859.harriet@bazleyfamily.co.uk>,
Harriet Bazley <harriet@bazleyfamily.co.uk> wrote:
Antony Sidwell's site http://ajps.mine.au and its Angband-Kit no
longer seem to exist.
[snip]
isparp1/angband/angband-kit.zipThe Wayback Machine search works on the Iris browser on RiscOS but
sadly the ajps site is reported as not archived.
That's not true !
It is archived - but the name ends with .nu (not .au)
At
<https://web.archive.org/web/20090130011516/http://ajps.mine.nu/ angband/ports/index.htm>
Nice detective work!
... but by crawling around and looking at where those download links
actually point, I can locate https://web.archive.org/web/20070126080114/http://homepage.ntlworld.com/
which *does* exist. That archive seems to be seriously mangled (the end
has been chopped off at some point, and there's no way of knowing what's missing), but using Infozip:zip -FF I can get a readable version of
what's left: version 1.31 of the porting kit as versus Musus' version
1.26. Apparently it features "some 32-bit cleanup by Harriet Bazley", which I have to say I don't have any memory of at all, and don't
remember ever being qualified to do, either!
By inspection of the actual zipfile I think it's probably only a few
bytes of the zipfile central directory that got truncated and not the
actual data as such, since the end of the archive consists of a list of filenames. In fact, by comparison of the download with my 'fixed'
version I think I can identify exactly where the file got cut off, which means there were 9432 bytes missing, all of which consist of filename
data duplicated in the actual body of the archive. So that means I've
now got an entire working copy, with nothing missing save for (by the
looks of it!) a certain amount of filetype data, which it should be
possible to deduce from context.
And it's got the vital main-ros/c file, which creates a functioning RISC
OS front end application.... (Having seen what happens with the SDL
version - no offence meant - I'd say that's fairly essential for getting
RISC OS users to accept the game.)
(Ah, I've just located my contribution: the BASIC source for the MuLite module, "reverse-engineered, bug-fixed and 32-bit-safed by H.Bazley 12th
May 2003"! I still have absolutely zero recollection of doing that,
but it is at least vaguely plausible that I might once have been
competent to do so.
All the calls to that module could probably be
dropped now, since all it does is supply *commands that didn't exist in
RISC OS 3.1 and 2.)
Right, I've been through setting all the filetypes by analogy with the
old version of the toolkit (and a little deduction), and I'm pretty sure
I've now got what ought to be a working version of the Angband porting
kit, with its complete instructions for cross-compilation etc. So
given the source for an Angband variant circa 2004, a C compiler
suitable for targeting the modern processors, and a sufficient knowledge
of C programming, makefiles, etc., it *should* be possible to recompile Angband for RISC OS and/or debug the cause of the various weird
errors...!
I can put it on my website, though it would be better hosted somewhere
else.
http://www.bazleyfamily.co.uk/angband/angband-kit.zip (217Kb)
The version I mentioned above and wich had been installed "here" is theI had version 3.0.9 from July 2007, which did fine... on the other
topmost - the "Red Dragon" one. This one did fine - ages ago.
computer! On the ARMX6 it just gives errors with the terrain definition files on the Cortex A9 processor.
The port on this page (version 3.0.6) crashes in a new and original way
with a "no writable memory at this address" error, and I have to type *Desktop to get back control. Aemulor makes no difference to either of
them.
Am Sun, 08 Aug 2021 21:15:18 +0100 schrieb Harriet Bazley:
I would also say that a little WIMP aroundings are extremly helpful in acceptance. The version I knew had only the "mainframe variant" of
save included, wich meant on had to use a save command from within the
games control commands. Since I tried v 3.0.5 yesterday the comfort
has been increased VERY by the inclusion of a save dialog an the
possibility to use F3 as the "normal" speedsave key. Extremley
helpful.
The other things like iconbar icon and a little menu with informations
about the authors are essential for RISCOS to get a good feeling. E.g.
Elite does this too - nobody needs this extra for the game, but its
much better integrated into the desktop experience. But the best thing
in 3.0.5 has been the option to maximize or at least "blow up" the
viewport of the game from its normal 40x24 (or so) to a much larger
view of the dungeon. This gives a massive better game overview. Short:
WIMP integeration is a must. And its done already.
I've seen the Mulite module and a ZAPreadraw module. The latter one can eventually simply be replaced by a newer variant wich is > ARM7/SA clean
and usable at the latest machines.
All the calls to that module could probably be
dropped now, since all it does is supply *commands that didn't exist in RISC OS 3.1 and 2.)
But it should be kept save - eventually - if(!) anybody can be found who
is willing to recompile the complete thing, it could be of much use if
there are some other compiles of newer angband versions >= v4 wich then
could be made run on older machines too.
Right, I've been through setting all the filetypes by analogy with the
old version of the toolkit (and a little deduction),
and I'm pretty sure I've now got what ought to be a working version
of the Angband porting kit, with its complete instructions for cross-compilation etc. So given the source for an Angband variant
circa 2004, a C compiler suitable for targeting the modern
processors, and a sufficient knowledge of C programming, makefiles,
etc., it *should* be possible to recompile Angband for RISC OS
and/or debug the cause of the various weird errors...!
I can put it on my website, though it would be better hosted somewhere else.
http://www.bazleyfamily.co.uk/angband/angband-kit.zip (217Kb)
Together with the last version v 3.0.9 this would be a very good move
towards a "NEW" game for the new and forthcoming generations RISCOS computers.
<https://web.archive.org/web/20200221040230/http://rephial.org/ downloads/3.0/angband-3.0.9-src.tar.gz>
If You have an account at ROOL You eventually could do a new thread with links to the angband-kit.zip to v 3.0.9 and to the Quickband Sources.
With some good luck this will activate somebody who is willing to do the (little) conversions and the recompiling.
By the way - I used the Angband v 3.0.5 I wrote about on a !RPCEmu with RISCOS 3.7 where it could be used in all its flavours. And it took at
least 2 hours - then I died in the deeper dungeon at Level 3 beeing
beaten by two big an poisened spider; no Chance; all 37 healthpoints gone away in seconds. Too bad. :( .... ;)
But: the same thing didn't started in RISCOS 5.22 (with and without the
free Aeomulor module); the RPi test will be done sometimes, but I think it will give the same result.
Sysop: | Keyop |
---|---|
Location: | Huddersfield, West Yorkshire, UK |
Users: | 296 |
Nodes: | 16 (2 / 14) |
Uptime: | 48:21:35 |
Calls: | 6,648 |
Files: | 12,198 |
Messages: | 5,329,987 |