Incorrect clues: Eta-Lamda-Hash

[th122]

I'm running into a number of clue-related issues on both MacOS and Windows. To keep things easier to track, I'll dump each sample into its own issue.

• [ ] MacOS: 0.8.3.1 binary (watson-0.8.3b.bin-macOSX.zip, identifies as 0.8.1 in "About" )
• [x] Windows: 0.8.4b binary (Watson.Installer.msi, identifies as 0.8.3 in "About" )

Eta < Lamda < Hash

Judged by the solution, 𝝶 was excluded correctly. The state of the puzzle doesn't allow that deduction though, and the reasoning given in the hint doesn't make sense to me.

[koro-xx]

Yep, this one is weird. It's evident that there's a bug in the clue system but it seems pretty random, I couldn't replicate it so far. I'll go through the logic part of the code and try to figure it out, but this one is tricky...

[th122]

I do have the savefiles for all errors I've encountered so far, but I'm struggling to upload them to github. I'll keep experimenting if another browser/OS combination will eventually allow me to attach them, hoping it would make the situations reproducible. (A Chromium-based browser did fare better, so here it is)

Watson.sav.gz

[koro-xx]

This one is certainly a bug. I will try to use the savefile to figure out what's going on, thanks!

[koro-xx]

Ah! That was helpful. It's a very stupid mistake on my part. I'll fix it but it may take me a little while (I have to recall how my obfuscated code works)

[th122]

Thanks a million, it's appreciated and still gaining fans (I got hooked quite recently by a former colleague, who managed to clock 40k+ games on the original Sherlock game still running in an emulator :)

Should I continue to fold the screenshots and savefiles into issues, or rather wait and check if your fix catches the other ones as well?

[koro-xx]

I think that won't be necessary (unless you see a bug of a different kind). I think I fixed it, but I can't make an installer from this laptop (i'm abroad at the moment). If you compiled from the sources, you should be able to pull the latest version and compile to test it out.

Otherwise, I made a debug version that you can test, but I don't even know if it will run since I have shared libraries for allegro. If you try it, let me know if it runs properly. Download here.

[th122]

I'm thoroughly impressed - that was fast! The download link said "Not found", so maybe there's a typo in the link? I´ll try if I can compile the sources - neither MacOS nor Windows are developing platforms for me, and the MacOS release is so long out of support that Homebrew compilation (needed for the allegro 5 libraries) is probably hopelessly broken, (dependency hell...). I'll give it a try, though.

[koro-xx]

Sorry, i forgot to release it. I fixed the url: here It's a zip file, it should run directly from the folder where it was extracted.

[th122]

Thanks, the debug release runs smoothly after adding the statically linked version of the required dlls to the directory/library search path. I fed my savegames to the debug release, and the suggestions for this one resolve beautiully, as does another with a similar problem, great!

Those savegames where the clues generated for solving the puzzle contained an error it cannot resolve either, obviously (they can only be solved by ignoring the erroneous clue) - see issue #7.

I haven't played enough new games to see if it generates erroneous clues, but I'll keep on trying :)

[koro-xx]

Would you mind letting me know which dlls you needed to make it work (so I can update the zip file)?. (on my laptop it worked with the libs I put together in the zip, even after removing the shared libs from other places, so I'm surprised it doesn't work elsewhere)

[th122]

The missing dlls it explicitely complained about were

• [ ] allegro_acodec-5.2.dll,
• [ ] allegro_audio-5.2.dll,
• [ ] allegro-5.2.dll,
• [ ] allegro_color-5.2.dll
• [ ] allegro_font-5.2.dll
• [ ] allegro_memfile-5.2.dll
• [ ] allegro_primitives-5.2.dll
• [ ] allegro_image-5.2.dll
• [ ] allegro_ttf-5.2.dll

I copied those from the bin directory of the allegro-x86_64-w64-mingw32-gcc-12.1.0-posix-seh-static-5.2.8.0.zip archive. Tracing back the error messages, those are all that are really needed by today's debug release (the one fixing #3 and #7 as well). In addition, watson.exe also needed executable file permissions, and in the "about" it identifies as v0.8.3 as of 2016 :)

[koro-xx]

yeah the versioning is a mess. Fixed that for now. Thanks for the info, I clearly don't know how linking works in windows (I didn't figure the dlls would be in the bin dir). Anyway, removing my mingw bin dir I can check properly for dependencies now. Besides those you mentioned I had to add libgcc_s_seh-1.dll , libstdc++-6.dll and libwinpthread-1.dll and now everything works fine. I repacked the zip file with the proper dlls.

[th122]

Thanks! Those libraries it only complained about when I tried the dynamically linked archive - with the statically linked allegro archive, it didn't complain about them. I'm pretty clueless when it comes to windows, but the cygwin ldd command was helpful to find out which libraries were used even on Windows. Didn´t expect that, typed it more out of habit :)

[th122]

The current version refuses to start with an error 0xc0000022.

[koro-xx]

Damn! I give up for now. I'm improvising here, I'll be able to put things together properly next week when I'm back home. Maybe try replacing the dll's I added with yours, or even removing the three new dll's I added (I think it will work since you have mingw64 installed, but probably won't work if you don't)

[koro-xx]

Could you check if the latest version with the installer runs properly?