Testing need for b90

[Azhrei]

Sorry, wolph, I had missed your posting somehow. I finally have a Win7 VM so I'll play with this a little myself and see if I get the same problems.

I do notice that you found the memory size setting and changed it (from the bottom of your post) but that still doesn't explain some of the earlier problems you found.

Thanks everyone for trying this out!

[Azhrei]

Hi Azh, just found time to run a test on my laptop. Previous install is b87 (twice, one with a special jar and one without),

I downloaded mtb90.zip, unpacked to desktop in a dir. and ran the launcher-130725.01.jar
This rendered the following error:
c:\..\mt.cfg line 1 incorrect and line 4 incorrect.

Strange. The default mt.cfg file had blank lines for lines 1 and 4. They're not supposed to be there and the launcher treated them as errors/warnings and displayed a message.

There's another issue in that the ~/.maptool directory isn't created if it doesn't already exist. I'll change it so the existence of that directory isn't required until the user presses the Launch button. That way they don't receive a "create it if necessary?" prompt until they're ready to run MT.

Next error I got 'windows cannot find -Xms64M' make sure....etc.

This one I couldn't reproduce but I only have the 32-bit J7 installed.

If you can make this happen again I'd like to see the full error message and the contents of the command line text field at the bottom of the launcher window.

error msg

TinyLaF v1.4.0
Theme: file:/C:/Users/Wolph/Desktop/MT%20B90%20beta/Default.theme
Aug 06, 2013 10:41:42 PM java.util.prefs.WindowsPreferences <init>
WARNING: Could not open/create prefs root node Software\JavaSoft\Prefs at root 0
x80000002. Windows RegCreateKeyEx(...) returned error code 5.

, but which does not interfere with mt.

This appears to be a JVM issue on Windows. See this page for a solution. (Google found that.) And this page for the Java bug report (note that it ends with "closed, will not fix" because they consider this a Windows issue).

It didn't, at least, the screen remained stuck on 'loading campaign' for quite some while, so I restarted....

Low memory.

same error in console and it didn't start, It remained stuck on a white screen saying 'grasslands' with the mt symbol. so quit again

Low memory.

and this time checked the memory (task manager), but not redundant java was running. So Started again. This time the launcher didn't run at all. So I guess an overall crash. Need to reboot and as im typing this while testing... ill submit this message and edit later on.

Yep, these are all low memory symptoms on Windows.

Skipping the rest of the (apparently) memory-related issues.

So I gathered that you;l need my debug info as well:

==== MapTool Information ====
MapTool Version: 1.3.b90
MapTool Home...: C:\Users\Wolph\.maptool
Max mem avail..: 227M
Max mem used...: 228M
Custom Property: -DMAPTOOL_DATADIR=C:\Users\Wolph\.maptool

Yep, you can see the low memory problem right there!

I noticed it said no IGD's though. Did you not have your router turned on at that time? Your other Help>Debug output did show it working correctly, I think.

... Hence I had a better look at the launcher and noticed that max mem per default is set to 256mb... changed that to 1024mb and tried again. That worked.

played around a bit and everything appeared to be working.

HTH

Yep, that's great! I'll fix the bug with the launcher not creating the directory if necessary (it wants to copy the logging XML files there, hence the messages you were getting) and then upload a new build with an announcement in this thread. But it'll likely be a couple days before I get to that point.

[wolph42]

- 'recreate issue': i'll try was a couple of weeks ago IRC I fully set the path to the java bin and that resulted in the message, resetting it to default fixed it (or the other way round)
- low memory: yes in the end I got there... I was wondering: isn't if more practical to set the default to 1024 its by far the MOST mentioned number I on this forum in relation to memory settings and IRC if your pc doesn't pull it you immediately get an memory error, while in my described case a user if confronted with 'broken' campaign files (at least thats how it most likely will be perceived).
- IGD: I was in spain at the time using someone else's internet connection, I have *no* idea what was used, so this I can't check for you, I *can* say however that it works at home.

edit: bigger issue:

New campaign file, drag 2 'clean' tokens onto the map and select them

If I run this code

Spoiler

Code: Select all

[h:currentSelection    = getSelected("json")]
[h:jsonOrder        = "[]"]
[h,foreach(tok, currentSelection), CODE:{
    [jsonOrder        = json.append(jsonOrder, json.set("{}","id", tok, "order", getTokenDrawOrder(tok)))]
}]
[h:broadcast(jsonOrder)]
<br>
[r:jsonOrder        = json.sort(jsonOrder,"a","order")]
 

(you can copy paste it straight into the chat) in b87 and b88 then it works fine, if I run it in b90-test it generates an error:

   Error in body of roll.       Statement options (if any): r       Statement Body : jsonOrder = json.sort(jsonOrder,"a","order")

it appears that json.order is broken in b90 test.

edit2: here a faster test:

Code: Select all

[r:jsonOrder        = json.sort('[{"id":"a","order":1},{"id":"b","order":2}]',"a","order")] 

just copy paste into chat.

[Azhrei]

Strange. Rev 5945 has a change (from 5942) in which the parameters to json.sort() past the second one are checked to see if they're JSON objects and throws an error if they're not. That's clearly wrong -- they're supposed to be strings.

Unfortunately, the commit note for 5945 is generic; it was when I updated the license text in every source file and apparently some changes that should have been separately committed were not. So I'll never know why that change was made, but I'm going to put it back the way it was.

Thanks!

Also, the bugs in the launcher are fixed and I'll put a new one in my DropBox if you care to check it out. My air conditioning is out right now and won't be fixed for a couple days, but I'm going to push hard to get Lee's undo/redo fix into b90 and then release it...

[Xaelvaen]

Just tested out your latest launcher (12 hours old, according to your dropbox) and it works well enough to open MT and run my stack-6 framework. I also tried the launcher packaged with your b90 (that one provided an error however, the 12-hour-old one did not). Regardless of this error, MT still functioned.

What I did notice, is that the shortcut keys for macros are no longer working (press F5 to adjust hit points, in example). I tried going into the macro, reassigning a function key, and trying again - still no result. The macros function when clicked, but alas, they cannot be hotkeyed.

Specifications; Windows Vista Machine, Java 6.x (32bit).

At this point, I moved on to testing b90 without the launcher - the same error occurs.

Final point, I'd like to mention that the X-ing out of a menu trick works for built-in menus, such as the Macro editor, but not for custom made frames. All of the frames I use for my campaign keep spawning in the center of the screen, no matter how many different ways I try to save their positions. However, the Chat/Initiative windows are staying put using the X-trick.

[wolph42]

tested the launcher and works flawlessly. Though I still am in favour of increasing maxmem to 1024 (or at lest 512). 256 is heavily outdated.

[Full Bleed]

tested the launcher and works flawlessly. Though I still am in favour of increasing maxmem to 1024 (or at lest 512). 256 is heavily outdated.

I concur. 256 is far more likely to cause problems than avoid them at this point.

How about adding something that sets the default to 512, but "suggests" 1024?

[Azhrei]

Just tested out your latest launcher (12 hours old, according to your dropbox)

Great, thanks for testing.

What I did notice, is that the shortcut keys for macros are no longer working (press F5 to adjust hit points, in example). I tried going into the macro, reassigning a function key, and trying again - still no result. [...]

Specifications; Windows Vista Machine, Java 6.x (32bit).

This is the first I've heard of this. Anyone else having an issue with this?

Final point, I'd like to mention that the X-ing out of a menu trick works for built-in menus, such as the Macro editor, but not for custom made frames. All of the frames I use for my campaign keep spawning in the center of the screen, no matter how many different ways I try to save their positions. However, the Chat/Initiative windows are staying put using the X-trick.

Yeah, I'm not exactly sure what is happening here. I do have a Windows VM set up for developing now (well, almost set up ) as well as one for Linux. I'm going to try this out on different platforms and see what I can find out.

tested the launcher and works flawlessly. Though I still am in favour of increasing maxmem to 1024 (or at lest 512). 256 is heavily outdated.

I concur. 256 is far more likely to cause problems than avoid them at this point.

How about adding something that sets the default to 512, but "suggests" 1024?

Okay, changing the default to 512 is easy enough. I hesitate to "suggest" anything since that means looking at the current machine's hardware config to determine how much memory is actually in it -- we don't want to suggest 1024 if there's only 1GB in the box to begin with.

But I can at least modify the tooltip so that it says something along the lines of, "Large maps and/or campaign files may require more memory. See forums.rptools.net for more information." (Hm, tooltips can't be clicked so I can't put an actual link in there, but the domain name might be enough.)

[Full Bleed]

Okay, changing the default to 512 is easy enough. I hesitate to "suggest" anything since that means looking at the current machine's hardware config to determine how much memory is actually in it -- we don't want to suggest 1024 if there's only 1GB in the box to begin with.

True. That's why I went with the safer 512k bet, even though any system with 2GB or higher should be going with 1024. I guess you can't pull the system mem and suggest against it?

Most machines less than 5 years old should have at least 2GB... unless we're talking about phones and tablets which might become more of an option in the future with the Android work-arounds being posted...

[Xaelvaen]

I still use my computer from 2000 to run Maptools for some of my house guest XD Back in the day, it was top of the line with the first ever 1gig processor and 512 RAM... Silly technology. Anywho, after some upgrades, it runs the old MT just fine.

To the point, however, I tried some isolation techniques Az (for the hotkeys not functioning) - tonight I'll try upgrading to Java 7 temporarily to see if that fixes the situation and let you know.

[Azhrei]

I guess you can't pull the system mem and suggest against it?

Correct. The Java strategy is "black box" in regards to the system it runs on, so hardware information other than which graphics resolutions are supported is hard to come by.

To the point, however, I tried some isolation techniques Az (for the hotkeys not functioning) - tonight I'll try upgrading to Java 7 temporarily to see if that fixes the situation and let you know.

That would be great. The site was down again, so I realize you've probably been frustrated for 24 hours or so.

[Xaelvaen]

Oh no worries at all - no frustration, I enjoy helping however I can. I'm horrible at coding - honestly surprised I picked up MTscript at all. So whatever I can do, y'know? I'm sure its just java 6.x +vista in some arrangement, cause b89 still works fine - so no frustration at all mate.

[wolph42]

just tested the hotkey F2 in b90 test on
campaign
token (on map and on other map, selected and impersonated)
lib:token

they all worked.

[Xaelvaen]

Finally had a chance to try out Java 7 and the hotkeys are still not working. They work just fine on both b87 and b89, so I ran the diag-

To be absolutely certain, I also just opened a blank campaign, made a simple 'testing' macro that just used [h: broadcast("Testing")] with the hotkey of F5 - no avail.

Hidden

==== MapTool Information ====
MapTool Version: 1.3.b90
MapTool Home...: ..\..\..\.maptool
Max mem avail..: 989M
Max mem used...: 66M
Custom Property: -DMAPTOOL_DATADIR=..\..\..\.maptool

==== Java Information ====
Java Vendor.: Oracle Corporation
Java Home...: C:\Program Files\Java\jre7
Java Version: 1.7.0_25
Result of executing 'java -version':
............: Java(TM) SE Runtime Environment (build 1.7.0_25-b17)
............: Java HotSpot(TM) Client VM (build 23.25-b01, mixed mode, sharing)


==== OS Information ====
OS Name........: Windows Vista
OS Version.....: 6.0
OS Architecture: x86

So yeah, it must be Vista I'm guessing =/

[Azhrei]

Hm. If they work with b87 and b89, that points to an issue with b90, not with the OS.

There was a change with regard to function key handing, IIRC. I'll try to find that patch in the ChangeLog and take a look at the code.

It could also be that a third-party library is being used in an incompatible manner in b90, but that seems very unlikely -- I don't think we use any 3rd party libs that are related to keyboard handling.