Re: [boost] Plea for addition of stream_cast

From: Kevlin Henney (kevlin_at_[hidden])
Date: 2000-06-29 15:59:13

In message <00eb01bfe207$73061490$056ca8c0_at_[hidden]>, David
Abrahams <abrahams_at_[hidden]> writes
>This runs into the same problem that your other "single-thread only"
>proposal has. Namely, if streaming A or C relied on the throwing behavior of
>an internal interpret_cast, that behavior would have  been changed
>externally.

The problems I was thinking of are quite different: in the first case
there was a problem with re-entrancy in the non-threaded situation, and
in the second the issue is that we do not have a portable way of
defining thread-specific data.

However, the implied stacked, global context itself is either a strength
or a weakness for the reasons that you mention :-)
____________________________________________________________

  Kevlin Henney                   phone:  [phone]
  Curbralan Ltd                   mobile: [phone]
  kevlin_at_[hidden]            fax:    [phone]
____________________________________________________________