Dmitry Timoshkov wrote:
"Shachar Shemesh" wine-devel@shemesh.biz wrote:
Ignoring LC_CTYPE is as random as ignoring any other one. For the 100th time, we are not ignoring LC_ALL and LANG. We are simply ALSO looking at LC_CTYPE.
We were looking for LC_CTYPE before and it was removed from the list of environment variables for a reason.
Yes. That reason was that it was setting both user and system locale. Wrong fix, through right direction.
It only appears unfortunately.
Then please qualify your statement. Where do they diverge?
That's nonsense to ask me or somebody else to proof your patch is wrong. That's your responsibility to proof its correctness to everyone else, writing test cases and pointing out to documentation if necessary.
When I pointed to documentation, you claimed (with no supporting evidence, I might add) that system and user locale are the same, and that the documentation is wrong. You since amended your claim, no longer claiming that system and user locale are the same, but still claiming the documentation is wrong. You now ask me to write tests proving the documentation is right? I'm sorry, I'm not going to through that insanity again.
In other words, this patch fixes Wine on my system without breaking it on yours. Any system on which it does break is a system you categorically call broken to begin with (LC_CTYPE points elsewhere from LANG). Why do you object so much?
I'm doing a lot of work supporting cxoffice users fixing locale related bugs for them. Since the release 3.0 I haven't had a single locale related bug report, and therefore I assume that most of the bugs are fixed.
That is not true. I have personally seen at least one message on the cx support mailing list where a CW employee (I think it was you) explained to an Israeli user (who was not me, mind you :) that using LC_CTYPE was wrong, and that their system is broken. Their system was working perfectly with Unix applications. As I have explained before, I reject that notion. Agree me with me or don't, you cannot claim that CX has no locale related problems.
Also, please bear in mind that crossover didn't support BiDi (at least, until I did http://www.lingnu.com/support.html), so a large part of the people potentially affected by this problem did not run it.
Now you want to break that fragile balance. How can I keep silence?
Dmitry, please believe me that I do not do things just so you have more work. However, you have to accept that your knowledge of the way Windows handles locales was flawed until yesterday. You also have to accept that you don't know all setups. As such, please consider the possibility that removing LC_CTYPE as something that initialize the user locale was an improvement, while introducing it as something that initialize the system locale is another improvement (rather than a set back).
Now, if you want to help, please compile a short list of the locale settings that caused you grief in the past, and let's go over them and see how Wine should handle them. This will allow us to move forward in a constructive manner.
Shachar