Forgot to add, it also gives these errors when trying to run a game with them on.

would be nice to know what the XXXXX is ?

also, see if it saved correctly by checking the ini

The X's are in hexadecimal. Here is an INI file that can be examined. It looks like it's saved correctly.

Maybe due to mth's recent changes?

Comment 4 by project member delroth@gmail.com, Today (13 hours ago)

"Maybe due to mth's recent changes?"¶

YES...I think...
I have just test 3.0-229 & 3.0-228...
In 3.0-228,AR codes works fine...(test games:SB3J08 - BASARA[Sengoku Basara 3] & S5QJC8 - SengokuMusou 3 Moushouden)
In 3.0-229,AR codes does not work,and codes doesn't show in AR codes list...
Same in 232&235...

r3.0-229 is Revision 27bda2c054f3

Could you try reverting only the "value == ULONG_MAX &&" part of that commit?

Also, does this bug occur on all platforms or only on Windows? And on 64-bit only or 32-bit as well?

I tried to reproduce the problem on 32-bit and 64-bit OS X, but the AR codes showed up fine there. So I think the issue is Windows specific.

This bit from MSDN's description of strtoul is a bit worrying: "The LC_NUMERIC category setting of the current locale determines recognition of the radix character in nptr; for more information, see setlocale." It means that the format of a hex number could differ depending on what locale settings you are using. Although it is unlikely to explain this issue, since the old code would have had the same problem.

forgot to say...my OS:win7-64
test both 64-bit and 32-bit (229 & 235)

I usually copy old config (Dolphin.ini...) to new ver.,in this way,it doesn't show the error like marioman...@gmail.com says...,can play the game,but AR codes just not work...

It shows error if not copying old config to new ver.
"invalid AR Code line: XXXXXXXX XXXXXXXX"
"The address is invalid"
"The value is invalid"

myuhh: Just opening the properties gives the errors here with the latest builds, not even trying to use an AR code. Windows 7, x64. This should be reverted until it gets sorted out.

Er, was supposed to say mthuu: :p

I'm not in favor of reverting because then the whole issue is likely to be forgotten. I'd like to get this fixed soon, but I need help with that since I don't have a Windows system at my disposal.

Isn't there a Windows coder out there who could have a look at this? The messages are PanicAlerts from ActionReplay::LoadCodes(). Apparently both the address and the value are being rejected. There are 3 "return false" statements in TryParse(), so it would be very useful to know which one of them triggers and on what input(s).

Actually, anyone who is capable of building Dolphin on Windows could try out the suggestion from comment #8, which would narrow things down quite a bit already.

mthuu: Master is not the place to try out code and hope that it works. It causes issues and should be reverted. Maybe next time you should commit to a branch and not subject everyone to code that causes issues.

For what it's worth, I'm on irc most days and can test your changes, it won't be forgotten if you fix it. I'll be the guy named [SS] and you can ask me to test your attempted fixes. Until then, please revert.

It may seem somewhat like a PC developer publishing a game without testing it on Ati(Amd) and/or Nvidia video cards and then asking the users to fix the bugs:-)

@#17: It wasn't a "let's hope it work" type commit: I checked the code, I wrote a unit test for it and cotton did the same on Windows. Somehow this issue slipped through, but it was not because of carelessness. I didn't revert it because I expected to be able to fix it very soon, but I need help for that from a Windows user and I've received no help yet. I hope we can sort it out on IRC today.

@#18: PC game devs have internal builds fail like that regularly, you just don't see those because they are tested and the issue is fixed before release. Devs and power users are our equivalent of a company's QA team, users who download only releases are the equivalent of users of PC games.

Thanks to skidau for testing and fixing this!

Thanks indeed.