In article <42574B24.7ECD6280@web.de>, Alexander Terekhov <terekhov@web.de> wrote:
You've (again) got it ass backwards, Miro. My employer should better provide some *reason* to follow that "strong recommendation" or withdraw it (preliminary status aside for a moment). Weak or strong, (preliminary) recommendation does NOT equate to erratum. See also atomic stuff for power* in glibc's sources and pay a little attention to '@...'.
Referring to glibc is not useful unless you give us a reason to believe that glibc was written with specific knowledge about why the recommendation should be ignored. For all we know, glibc's reason for being the way it is is the same as Peter's was -- i.e., not knowing about the IBM notice. Given that you work for IBM, you may have access to someone who knows why this preliminary notice was issued and why it is safe to ignore it, if it indeed is. If you are an authority on this issue, then please explain why the preliminary notice was issued and why it's safe to ignore it. If you are not an authority, then find someone who is and let us have the explanation. Otherwise, you are doing nothing but repeating poorly substantiated claims and urging us take your word for it. Finally, I would appreciate it if you could back off from your condescension. My goal here is not to insult you, prove you wrong, or make you look like a fool. My goal is to do what I can to ensure that when I build boost into a product and ship it to tens of thousands of users, that product's reliability and future compatibility are not compromised by poor engineering. meeroh