Accelerator keys are not used consistently

[jeffr]

There are a number of places where libpurple returns "OK" instead of "_OK", for example. These should all be made consistent.

[deryni]

A listing of places or a patch to fix them would be much appreciated.

[jeffr]

Leave it to me :)

[bernmeister]

jeffr: Any news on this...?

[darkrain42]

All from im.pidgin.pidgin at 18f16a423f835272cdecfadd92ac25f7dfa9ce50

gtkcertmgr.c:195:				     _("OK"),
gtkdialogs.c:754:						_("OK"), G_CALLBACK(pidgin_dialogs_im_cb),
gtkdialogs.c:893:						_("OK"), G_CALLBACK(pidgin_dialogs_info_cb),
gtkdialogs.c:985:						_("OK"), G_CALLBACK(pidgin_dialogs_log_cb),
gtkrequest.c:272:	STOCK_ITEMIZE(_("OK"),      GTK_STOCK_OK);
gtkutils.c:1604:						    DND_FILE_TRANSFER, "OK", (GCallback)dnd_image_ok_callback,
gtkutils.c:1624:							"OK", (GCallback)dnd_image_ok_callback,

account.c:1158:                        _("OK"), ok_cb,
account.c:1493:						_("OK"), G_CALLBACK(change_password_cb),

[deryni]

Those appear to all be in purple_request_* calls, which translate the text to a Stock ID (that's where the gtkrequest.c line comes from). As such those should all get the correct mnemonics set for them, except for the un-localized versions as the stock-id matching checks localized strings. Which leads me to believe that we should actually stop localizing the text and probably add PURPLE_STOCK_{OK,CANCEL} defines to make this clearer going forward. Seem reasonable?