Transport Tycoon Forums

Audiopulse wrote: May I suggest just linking it in the OP?

Audiopulse wrote: Edit2:
I noticed some graphical glitches - but nothing else impacting gameplay, really. OpenMars is quite playable and Im having great fun.

Ill post a few Screenshots tomorrow showing offsets and misalignments I found.

Audiopulse wrote:I might just do that, planetmaker. Like I said - there are a few crazy ideas I would like to give a try with this GFX ... I have a wild fascination with space I have been tending to since being a kid

Right now I am not at home, so this would have to wait another 2 weeks, maybe - what do you think, would zephyris be okay with me changing or adding new things?

Imgur-gallery with everything i noticed.

Audiopulse wrote:What I know about Repositories, Forks and Hubs I know from hearsay - but Ill get back to you as soon as a need arises. Thank you!

I would have to start by taking the GRF apart and look at things, anyway, so it'll be some time anyways

planetmaker wrote: 05 Feb 2020 16:37 Both URLs are actually the same thing just with a different web interface.

The individual URLs seem to work fine (and can be cloned separately and they all work separately), yet the parent repo which exists for convenience and contains all repos shows the error you mentioned. I wonder why... and shall investigate. Thanks for letting me know.

ToxicFrog wrote: 06 Feb 2020 00:37 Thank you! Hopefully it's not too difficult to fix; I don't know much about hg, just the very basics.

I'm not sure what's going on with these forums; the post I made this morning with instructions on how to clone the repos (which planetmaker is responding to) has evaporated and my account got deleted as well. I had to re-register. I can re-post them if anyone needs, but hopefully it'll be an easy fix for planetmaker and people will be able to just clone the top-level repo again.

hg clone https://hg.openttdcoop.org/opengfx-mars

$ cat .hgsub
(...)
ssh://hg@hg.openttdcoop.org/opengfx-mars/graphics   = ssh://hg@hg.openttdcoop.org/opengfx-mars-graphics
(...)
http://hg.openttdcoop.org/opengfx-mars/graphics   = http://hg.openttdcoop.org/opengfx-mars-graphics

planetmaker wrote: 06 Feb 2020 11:21 Thank you for being that persistent and re-registering. I don't have enough insight into forum logs to know what happened exactly... the only explanation I see is human error in spammer removal. However that happend, sorry

On-topic: I think I found the root of the issue: too many URL aliases, and only the original repo URL is taken care of.

[...]

EDIT: I pushed an appropriate commit, cloning now works for me for http://kallithea.openttdcoop.org/opengfx-mars as well

ToxicFrog wrote: 06 Feb 2020 12:53 On testing, it looks like you need to rewrite http and https separately, though -- cloning http://hg.openttdcoop.org/opengfx-mars works ok, but cloning https://hg.openttdcoop.org/opengfx-mars doesn't. Same deal with kallithea. Looking at .hgsub there are only rules for http. Since the web interface gives https links, it would probably be good to have both.

planetmaker wrote: 06 Feb 2020 13:08 I noticed that some of the sub-repos will need some changes to actually build - at least with the toolchain I currently have on this machine. Probably not a big deal, but will need doing before making changes, I guess. Feel free to make pull-requests... or request repository access.

• SHELL=/bin/bash doesn't work on NixOS. This is probably not an issue on other linuxes, though, and can be fixed on the command line by invoking with `make SHELL=$(whence bash)`, so it's likely not worth fixing.
• findversion.sh uses "python", but uses a py2-only API, so it fails on modern systems where "python" is py3; replacing it with an explicit call to python2 fixes it, thus:

  Code: Select all

  sed -E -i 's,python -c,python2 -c,' findversion.sh */findversion.sh

• Some of the Makefiles have NML ?= "$(shell which nmlc 2>/dev/null)"; this is bad because if nmlc is in PATH it's just a more roundabout way of doing NML ?= nmlc (which some of the other Makefiles do already), and if it's not the enclosing quotes break some error checking later in the file! At minimum the quotes need to be removed. Quick sed hack:

  Code: Select all

  sed -E -i 's,"\$\(shell which nmlc 2>/dev/null\)",nmlc,g' Makefile */Makefile

sed -E -i 's,^UNIX2DOS_FLAGS,UNIX2DOS_FLAGS ?= -q,' Makefile */Makefile

ToxicFrog wrote: 07 Feb 2020 01:07 SHELL=/bin/bash doesn't work on NixOS. This is probably not an issue on other linuxes, though, and can be fixed on the command line by invoking with `make SHELL=$(whence bash)`, so it's likely not worth fixing.

ToxicFrog wrote: 07 Feb 2020 01:07 [*] findversion.sh uses "python", but uses a py2-only API, so it fails on modern systems where "python" is py3; replacing it with an explicit call to python2 fixes it, thus:

Code: Select all

sed -E -i 's,python -c,python2 -c,' findversion.sh */findversion.sh

ToxicFrog wrote: 07 Feb 2020 01:07 [*] Some of the Makefiles have NML ?= "$(shell which nmlc 2>/dev/null)"; this is bad because if nmlc is in PATH it's just a more roundabout way of doing NML ?= nmlc (which some of the other Makefiles do already), and if it's not the enclosing quotes break some error checking later in the file! At minimum the quotes need to be removed. Quick sed hack:

Code: Select all

sed -E -i 's,"\$\(shell which nmlc 2>/dev/null\)",nmlc,g' Makefile */Makefile

ToxicFrog wrote: 07 Feb 2020 01:07

Code: Select all

sed -E -i 's,^UNIX2DOS_FLAGS,UNIX2DOS_FLAGS ?= -q,' Makefile */Makefile

This will fail on any system where unix2dos does not accept "-q" but at least works on systems where it does. A better fix might be to replace the 2>/dev/null with >/dev/null 2>&1.

ToxicFrog wrote: 07 Feb 2020 01:07 With that change made `make` runs to completion and spits out a bunch of GRFs; hopefully that'll be enough to get it working for you, too, although there seems to be a lot of fragile and not very portable stuff in these Makefiles. :/

planetmaker wrote: 07 Feb 2020 09:24 One could try whether /bin/sh works. But I'm reasonably confident that I implemented some bashisms into the shell commands. Though... making it a bit more portable definitely wouldn't hurt. It worked for me back then, it works for the build machine - that's what I tested it against, including the systems of other people using it.

I'd suggest a port to python3 instead.

Agreed. NML = ?= nmlc should do the trick. The problem with nml's PILLOW version sounds familiar and might already be fixed in the development versions. Which nmlc do you use?

I had the same problem yesterday and pushed a fix. But the main repo references still an older revision of the aircraft set. So update to the latest version of the aircraft set and you should see it fixed. (The main repo explicitly stores the revision for reach sub-repository to use - and I didn't update it yesterday).

I actually stopped yesterday when my nmlc complained about some properties in the industry set not being available... that needs some looking into how nml syntax changed... alas!

Recently I thought about whether 'stealing' the apollo industries set (released for last year's 50th anniversary) found somewhere here in the forums could be integrated somehow here... it would fit the theme. And on Mars you definitely want to have some space industry

ToxicFrog wrote: 07 Feb 2020 13:20 Huh. 0.4.5 builds all the GRFs for me, albeit with a bunch of warnings and I have not tried loading the generated GRFs to see if they work.

Recently I thought about whether 'stealing' the apollo industries set (released for last year's 50th anniversary) found somewhere here in the forums could be integrated somehow here... it would fit the theme. And on Mars you definitely want to have some space industry

This? viewtopic.php?f=67&t=85580

That looks extremely cool, although it's not really aesthetically harmonious with the Mars graphics -- maybe Mars should have a landing pad for those rockets instead

planetmaker wrote: 07 Feb 2020 13:52 I have a build of nml's current master. The way industry production is handled changed between 0.4.x and current master / 0.5.0 pre-releases to accomodate the possibility to have 16 input and 16 output cargoes with the upcoming OpenTTD 1.10.0.

Yes, I meant exactly that NewGRF. I agree that the graphics don't fit too well, so some work on that, too, but it's a start to build on. Yet I very much like the concept Maybe landing pad additionally

planetmaker wrote: 07 Feb 2020 15:18 We moved the nml repository to github along where the other OpenTTD tools reside (now): https://github.com/OpenTTD/nml

Moving these repos to github probably would not be a bad idea, either... more people know git nowadays than do know hg (even though the entry barrier for hg still seems smaller to me).