Am 08.11.2012 09:38, schrieb Paul Isambert: > Selon Taco Hoekwater <[email protected]>: > >> On 11/07/2012 07:37 PM, Stephan Hennig wrote: >> > Am 06.07.2011 15:43, schrieb Taco Hoekwater: >> >> On 07/04/11 09:06, Paul Isambert wrote: >> >>> >> >>> The manual says you're in charge of ensuring that in >> >>> "node.insert_after(H, N, x)", N is in the list denoted by H. The thing >> >>> is, H doesn't seem to matter at all, unless it's nil: >> >> >> >> Actually H is also useful if N is nil (that is a tail-append the list >> >> that starts at H), but it is true that H is generally unused. >> > >> > If N is known to be non-nil, e.g., a glyph node, is it save to call >> > >> > node.insert_after(nil, N, x) ? >> > >> > If the answer is 'Yes', >> >> Well, yes. However I do not like the idea of reordering the >> arguments, because we have quite a lot of functions with 'head' >> argument, and they always come first. > > I definitely agree (don't want to rewrite umpteen files).
Come-on! There might be umpteen TeX and LaTeX files, but no Lua code files that use node.insert_after. > Plus the optionality of the head makes less sense than the > optionality of the current node, even if it were more frequent; it'd > mean you have to check beforehand whether the current node exists or > not before you use the function. Well, in my use-case, current is always a valid node. So I wondered why head were needed at all. But your use-cases might differ. But then again, current has to be checked for being nil anyway (even if that is hidden in the function). > If the head is a mandatory argument, you just don't ask question. I think inserting a node at the end of a given head node is something different than inserting a node exactly after a given node. I you want to do the former, you could always say insert_after(node.tail(head), new) Best regards, Stephan Hennig
