30 Dec
2010
30 Dec
'10
5:09 a.m.
AMDG On 12/21/2010 12:29 PM, Jeffrey Lee Hellrung, Jr. wrote:
I would like to see a situation where finding boost::fusion::begin/end via ADL is actually useful (i.e., couldn't be replaced by explicit qualification). Until then, I believe preventing boost::fusion::begin/end (and, possibly, boost::begin/end) from being found via ADL is a better solution; further, it explicitly discourages any future code from relying on ADL (assuming that relying on ADL *should* be discouraged).
ADL for Fusion may be convenient in some cases, but it should never be necessary because fusion::begin/end are not the customization points. In Christ, Steven Watanabe