search

adventuregamestudio / ags

AGS editor and engine source code
Other
705 stars 159 forks source link

ags3.5: "Unable to find game data file" #808

Closed fernewelten closed 5 years ago

fernewelten commented 5 years ago

I'm building AGS ags3, newest version (commit e832a4fc). I call up the Editor for the program referenced below. Compiling it (f7) and clicking on Compiled/Windows/Test01.exe (or …/winsetup.exe) calls up a popup window with the message: “ERROR: Unable to find game data files.” Same effect when test-running it (f5) from the Editor.

(This program compiles and runs in AGS 3.4.1 without errors.) Test01.zip

ghost commented 5 years ago

There's something very strange here. I make Editor + Engine from same commit, and do same steps, yet the game runs well (I can see "Hello World" label with input field on screen).

I also noticed that compiled game is larger the the one in your archive (7 MB vs 5 MB), which maybe means that in your case some data was not appended (or different data was appended).

ghost commented 5 years ago

Checking the information on executables found in your archive in _Debug and Compiled/Windows, the version of the engine is 3.4.2.1, which is a very old dev version of 3.5.0. Have you rebuilt the engine from same commit also?

fernewelten commented 5 years ago

I'll carefully retry and rebuild the steps and report back. But generally, I went in the other direction: from top downwards:

I'm trying to secure a reliable release to base my live tests on.

After this “adventure” of a different kind, I reconstructed my steps to write up the error reports; I might have left out a recompiling step at that point in time.

ghost commented 5 years ago

Starting on master, I tried to construct a Hello World adventure and had the editor repeatably crash on me.

I have to warn that this bug #809 will happen to you regardless of which version you build. It's been in AGS since the beginning of 3.* generation. The workaround is to comment out the assertion. In fact, maybe it should be commented and commited to the branch to prevent same confusion, but I was never certain if this happens to other people because no one else mentioned it before (except myself)

So as a last resort, I looked up the exact version that was claimed to be “stable” on the AGS download page, and tried to get that to compile and run.

3.4.* branch was originally developed with MSVS 2008. We had to do number of steps upgrading it and making sure it compiles on MSVS 2015 and 17, but I guess these steps are not in the release-3.4.3 branch itself, but in next one (which is now known as ags3). These changes could include both fixes to the code syntax, and to project settings. Also automatic conversion may add mistakes on its own.

fernewelten commented 5 years ago

Okay, carefully recompiling everything makes the problem go away. I seem to be all set. Thanks!