Some Qs for James on the pakbritain-experimental repository....

Topic: Some Qs for James on the pakbritain-experimental repository.... (Read 9645 times) previous topic - next topic

Some Qs for James on the pakbritain-experimental repository....

April 17, 2010, 05:44:14 pm

I'm having trouble building it. This seems to be largely because the files are split between the top
directory and the Pak128.Britain-Experimental subdirectory, and the makeALL.mos script isn't set up for that.

Is it possible to just merge everything back into the top level?.... or is there a reason for this, and should I edit the MOSE script?

EDIT: Also, I'm getting a lot of this:
libpng warning: Invalid pHYs after IDAT

On, like, every single file.
EDIT: OK, not every one, just a whole lot of them.

Re: Some Qs for James on the pakbritain-experimental repository....

Reply #1 – April 17, 2010, 06:43:15 pm

Neroden,

ahh, I don't actually use the script to build it - I do it manually at the command line, so I'm very unfamiliar with the script, I'm afraid. The directory structure should be the same as for the Standard SVN, though.

Edit: Your compat.tab entries for the bridges are incorrect, incidentally: there needs to be just one new item for every old item specified (although each old item can be specified more than once). For example:

OldThing
NewThing_1
OldThing
NewThing_2

...and...

OldThing1
NewThing
OldThing2
NewThing

...and so forth.

Re: Some Qs for James on the pakbritain-experimental repository....

Reply #2 – April 17, 2010, 11:37:55 pm

Quote from: jamespetts – on April 17, 2010, 06:43:15 pm

Neroden,

ahh, I don't actually use the script to build it - I do it manually at the command line, so I'm very unfamiliar with the script, I'm afraid.

You know, this is probably why certain file changes are not getting into the pak -- you're forgetting to recompile them.

Quote

The directory structure should be the same as for the Standard SVN, though.

All right then! It wasn't, but pull from my layout-cleanup branch and it will be. I had to make some choices when faced with two competing variants of various files, however, so you'll want to take a look at the differences.

Quote

Edit: Your compat.tab entries for the bridges are incorrect, incidentally: there needs to be just one new item for every old item specified (although each old item can be specified more than once). For example:

[/quote]
Whaaa?

For each old item, it gets converted to wooden_road_bridge_wood. I'm collapsing three old items into one new item. I *think* I did this correctly. If I didn't, the format is *very* confusing.

Re: Some Qs for James on the pakbritain-experimental repository....

Reply #3 – April 18, 2010, 12:02:27 am

Neroden,

ahh, my error: I had misread your file: you have done it correctly. However, I think that you picked the wrong version of compat.tab to integrate: we need the following:

Code: [Select]

############################ Victorian coal mine ##############################
VictorianCoalMine
CoalMine1850
################################ Brick viaducts ################################
BrickViaduct175
BrickViaduct
BrickViaduct200
BrickViaduct
BrickViaduct225
ConcreteSupportedBridge
BrickViaduct320
ConcreteSupportedBridgeHighSpeed
############################## Stone road bridges ##############################
stone_road_bridge_dirt
MasonryViaductDirt
stone_road_bridge_cobbles
MasonryViaductCobbles

And I usually do recompile everything whenever I release a new version. I must confess, I'm not quite sure how to use the script, especially on Windows - any pointers?

Re: Some Qs for James on the pakbritain-experimental repository....

Reply #4 – April 18, 2010, 12:45:47 am

Quote from: jamespetts – on April 18, 2010, 12:02:27 am

Neroden,

ahh, my error: I had misread your file: you have done it correctly. However, I think that you picked the wrong version of compat.tab to integrate: we need the following:

Right, thanks, that was my error. The correct one was the one I used for the layout-cleanup branch....

Quote

And I usually do recompile everything whenever I release a new version.

I'm guessing a failure to recompile accounts for the wind ferries and barges not having any Waterway restrictions in 7.3, when they have them in the repository. And a recompiled version has them in the game. (Although my recompiled version is giving me weird crashes.)

Quote

I must confess, I'm not quite sure how to use the script, especially on Windows - any pointers?

Yes, it is rather tedious, isn't it? I haven't quite worked out all the details myself, but....

(1) You *need* Python 2.6 or later. If you haven't managed to install python, that's the first step.

http://www.python.org/download/releases/

http://docs.python.org/py3k/using/windows.html

(2) You need MOSE, which is available from this forum:
http://forum.simutrans.com/index.php?topic=2403.0

The magic incantation appears to be "python mose.py makeALL.mos". (Followed by "quit".)

I would not call it the best possible build system: http://forum.simutrans.com/index.php?topic=4578.0

But it is the one which is mostly there already. I could build a Makefile instead, but then you'd have to figure out how to install Make....

Re: Some Qs for James on the pakbritain-experimental repository....

Reply #5 – April 19, 2010, 05:35:54 pm

Let me explain what's going on with the repo layout.

There's a "top level" of the repo, which is laid out the same way as the SVN repo for pak128.britain (Standard).

Then there's a subdirectory called "Pak128.Britain-Experimental". Something like half of the files, entirely at random, are in this subdirectory rather than the main directory, in a copy of the same layout.

My "layout-cleanup" branch consisted of overlaying all the files from the the "Pak128.Britain-Experimental" tree onto the main directory tree, and then deleting the "Pak128.Britain-Experimental" subdirectory tree. I *STRONGLY* advise that you do this ASAP. The current layout is very confusing.

James, please pull from my layout-cleanup2 branch.

Reply #6 – April 21, 2010, 04:46:03 pm

As of today, my layout-cleanup2 branch has the same data as your master branch (for pak128.britain-experimental), but has a sane directory layout.

Please, please, please, please merge from it.

Re: Some Qs for James on the pakbritain-experimental repository....

Reply #7 – April 21, 2010, 11:23:23 pm

Neroden,

thank you for this work, and apologies that I have not looked into this sooner. Attempting to merge, however, I get this error:

Code: [Select]

Merging:
4bc5929 ADD: L&YR EMU
virtual Neroden-Pak128.Britian-Ex/layout-cleanup2
found 1 common ancestor(s):
a0f20dd FIX: Incorrect running cost for Morris 8 post van
error: Untracked working tree file 'text/en.tab' would be overwritten by merge.
fatal: merging of trees fdf39f44de48696afbae56091aee399573f3a834 and 823253b0595b5396901f176423df482110bb56de failed

Re: Some Qs for James on the pakbritain-experimental repository....

Reply #8 – April 22, 2010, 12:12:30 am

Quote from: jamespetts – on April 21, 2010, 11:23:23 pm

Neroden,

thank you for this work, and apologies that I have not looked into this sooner. Attempting to merge, however, I get this error:

Code: [Select]
error: Untracked working tree file 'text/en.tab' would be overwritten by merge.

That means: throw away your copy of text/en.tab and try again. Unless you actually have local modifications to text/en.tab. In which case, please commit them first, *then* merge.

Re: Some Qs for James on the pakbritain-experimental repository....

Reply #9 – April 24, 2010, 01:23:49 pm

Neroden,

as you will see, I have now merged your changes. However, after doing this, I can no longer merge changes from the standard Pak128.Britain SVN: when I run "SVN Update", I get the following error message:

Error Working copy 'F:\My Documents\Development\Simutrans\PakBritain sources\Experimental' locked. Please execute 'cleanup' command.

When I execute the "cleanup" command, I get this error message:

Cleanup failed to process the following paths:
-F:\My Documents\Development\Simutrans\PakBritain sources\Experimental
'F:\My Documents\Development\Simutrans\PakBritain sources\Experimental\text' is not a working copy directory.

Any ****istance in resolving these errors would be much appreciated.

Re: Some Qs for James on the pakbritain-experimental repository....

Reply #10 – April 24, 2010, 05:34:25 pm

Quote from: jamespetts – on April 24, 2010, 01:23:49 pm

Neroden,

as you will see, I have now merged your changes. However, after doing this, I can no longer merge changes from the standard Pak128.Britain SVN: when I run "SVN Update", I get the following error message:

Error Working copy 'F:\My Documents\Development\Simutrans\PakBritain sources\Experimental' locked. Please execute 'cleanup' command.

When I execute the "cleanup" command, I get this error message:

Cleanup failed to process the following paths:
-F:\My Documents\Development\Simutrans\PakBritain sources\Experimental
'F:\My Documents\Development\Simutrans\PakBritain sources\Experimental\text' is not a working copy directory.

Any ****istance in resolving these errors would be much appreciated.

I made two commits to layout-cleanup2 which may help. Or they may not, because there was something funny going on with your svn checkout in the first place (hence the confused directory layout).

I now know how you're syncing with SVN. :-P You have the same working directory for both svn and git, right? This... isn't such a great idea.

I am trying to set up a git repo tracking the Pak128.Britain SVN, like we have for the code for simutrans standard, which would allow this to be dealt with in a much cleaner way. This *almost* works trivially -- it ought to be easy enough that if I went away and stopped updating (always a real possibility), you could do the same thing using the "git-svn" instructions on github. (Noting that you need to use the -T option to git-svn, because the pak128.Britain layout is nonstandard.)

Unfortunately I'm getting hammered by line-ending format conflicts -- see my message to TheHood.

If the line endings are 'normalized' in SVN I think my problems will go away. Both SVN and git have a 'bias' towards a particular line ending storage style; if it's CRLF in the repository (as it is now in SVN), it will *never* play nice with Linux/Mac, but if it's LF in the repo, it will play nice with Windows.

However, you're on Windows, so it might possibly "just work right" for you. See if you can merge from the "Pak128.Britain" repo under my account at github.

Re: Some Qs for James on the pakbritain-experimental repository....

Reply #11 – April 26, 2010, 05:38:18 pm

Neroden,

thank you for your help with this. Unfortunately, your Git commits to layout-cleanup2 did not ****ist the SVN problem. However, you are probably right that the way to go is to become Git-exclusive if it is possible to set up an automatic pull from SVN and push to Git system as we have with the Standard code.

I have tried to set up your Pak128.Britain Standard repository, but get an odd error when I try to fetch from it:

Code: [Select]

walk df553fb1aac5d538feda44c801767c1e1c7dce66
walk 7adc2aced59593d2b3a586a2eb428336d736dd5a
Getting alternates list for http://github.com/neroden/pak128.Britain.git
Getting pack list for http://github.com/neroden/pak128.Britain.git
Getting pack 363885240a10482b09f971ed4080061a2ae48db0
 which contains 9ecae95026cc2edc5be233b463584dc20ed37d29
error: Unable to get pack file http://github.com/neroden/pak128.Britain.git/objects/pack/pack-363885240a10482b09f971ed4080061a2ae48db0.pack
transfer closed with 78607524 bytes remaining to read
error: Unable to find 9ecae95026cc2edc5be233b463584dc20ed37d29 under http://github.com/neroden/pak128.Britain.git
Cannot obtain needed tree 9ecae95026cc2edc5be233b463584dc20ed37d29
while processing commit 7adc2aced59593d2b3a586a2eb428336d736dd5a.
fatal: Fetch failed.

Might this be related to the line endings issue?

Re: Some Qs for James on the pakbritain-experimental repository....

Reply #12 – April 26, 2010, 06:36:46 pm

Quote from: jamespetts – on April 26, 2010, 05:38:18 pm

Neroden,

thank you for your help with this. Unfortunately, your Git commits to layout-cleanup2 did not ****ist the SVN problem. However, you are probably right that the way to go is to become Git-exclusive if it is possible to set up an automatic pull from SVN and push to Git system as we have with the Standard code.

I have tried to set up your Pak128.Britain Standard repository, but get an odd error when I try to fetch from it:

OK, that mystifies the heck out of me! I don't really understand git innards thouugh.

I have done some funny stuff with that repo, due to not really knowing what I was doing when I started. Could you try making your own clone of it into a fresh directory now and see if you get the same error (or a different one)?

If you get the same error I may try rebuilding the version on github, because my local version seems to be working OK.

If it checks out OK into a fresh directory, then it *is* a merge-related issue. :-(