Difference between revisions of "Talk:BIP 0014"
(→Arguments about requirements: new section) |
(→Arguments about requirements) |
||
| Line 8: | Line 8: | ||
| − | Luke has been arguing that clients are required to write specific information into the various fields and forbidden from lying e.g. a merchant daemon can't claim to be running a GUI in order to avoid making it a target for attack. I don't believe this is actually required by the BIP as written. Nor do I think requiring it would make sense: the requirement can't really be enforced, required information disclosure would make users less secure, and the expectation of certain uses would lead to programs changing their behavior based on it because they think they can depend on it (much like all-browsers-are-mozilla). I think this can be clarified by the judicious addition of MAY and SHOULD into the document. Unless there is opposition I will go ahead and do it. --[[User:Gmaxwell|Gmaxwell]] 04:08, 23 December 2011 (GMT) | + | Luke has been arguing that clients are required to write specific information into the various fields and forbidden from lying e.g. a merchant daemon can't claim to be running a GUI in order to avoid making it a target for attack. I don't believe this is actually required by the BIP as written. Nor do I think requiring it would make sense: the requirement can't really be enforced, required information disclosure would make users less secure, and the expectation of certain uses would lead to programs changing their behavior based on it because they think they can depend on it (much like all-browsers-are-mozilla). I think this can be clarified by the judicious addition of MAY and SHOULD into the document. Unless there is opposition I will go ahead and do it. I think the purpose of separating the useragent and the protocol is an excellent one, which the community clearly supported, but I don't think most people expected this to be some surprise legislation forcing nodes to identify themselves on the network in particular ways. --[[User:Gmaxwell|Gmaxwell]] 04:08, 23 December 2011 (GMT) |
Revision as of 04:13, 23 December 2011
Suggestion: be more explicit that the version in the protocol is a 32-bit integer, that 0.6.0 is represented by the number 60000, and that the 'subversion' field is an arbitrary string. --Gavin Andresen 19:19, 16 December 2011 (GMT)
Just as a point of information; "git describe" output solves the problem of non-incremental committish. It does this by providing a base version code (the most recent tag name) the number of commits past that tag and then a little bit of the commitish for the revision in question. --Andyparkins 11:45, 9 December 2011 (GMT)
Arguments about requirements
Luke has been arguing that clients are required to write specific information into the various fields and forbidden from lying e.g. a merchant daemon can't claim to be running a GUI in order to avoid making it a target for attack. I don't believe this is actually required by the BIP as written. Nor do I think requiring it would make sense: the requirement can't really be enforced, required information disclosure would make users less secure, and the expectation of certain uses would lead to programs changing their behavior based on it because they think they can depend on it (much like all-browsers-are-mozilla). I think this can be clarified by the judicious addition of MAY and SHOULD into the document. Unless there is opposition I will go ahead and do it. I think the purpose of separating the useragent and the protocol is an excellent one, which the community clearly supported, but I don't think most people expected this to be some surprise legislation forcing nodes to identify themselves on the network in particular ways. --Gmaxwell 04:08, 23 December 2011 (GMT)
