Hi Donal, thanks for that. Comments on your issues in-lined below. On 11 Sep 2008, at 18:29, Donal K. Fellows wrote:
[...]
Bugfix: inconsistent paragraph separation on page 3
Done.
Bugfix: example on page 6 uses DocumentNode to enclose XPath without any namespace mappings; suggested fix: note in text above that this is an abbreviated form for example purposes.
Added the necessary mappings to be consistent with the rest of the document.
Bugfix: example on page 8 has similar problems, and has "[sweep]" just above it (not a syntax recognized from elsewhere or defined here!) though it looks like a broken reference to another document?
Added the necessary mappings to the example. Replaced "[sweep]" with "Sweep".
Bugfix: page 9 contains many uses of the odd syntax described above.
Those notations refer to their definition in the XPath specification. Therefore I refrained from changing that syntax.
Bugfix: footnote 1 on page 9 contains multiple references to the same document!
Duplicates removed.
Bugfix: inconsistent paragraph separation on page 10
Fixed.
Query: is it worthwhile stating what happens with substring selection when using multiple assignments into the same field and some of those assignments change the length of the overall string? (Question brought about by reference to Example 3 (pp11-13) though that doesn't have this problem.)
I would argue that this is implementation dependant. I think this is not only a corner case but also a case where I doubt we will ever reach consensus on what the right way would be to behave in this case. I would therefore let the market decide in their implementations.
Bugfix: on too many pages to list, we're inconsistent about formatting of XML QNames and XPath fragments.
Fixed (completely, I hope). Were many spots to fix ...
Query: is it worth using an abstract mathematical description of sweep functions? (They're a pair, consisting of a maximal count and a function from an integer to the DOM node to substitute.)
I thought about this right from start. I would say nay for practical reasons. The document is already fairly abstract. Making it even more abstract would make it much more difficult to relate/translate to a real implementation, I guess.
Query: is it worth noting that implementations may wish to constrain the allowable space of values for LoopDouble/@step to exclude zero?
The description of the LoopInteger function gives an explicit value range that implementations must support. I fail to see a reason for implementations to further constrain that value range.
Bugfix: in Examples 7-8 (pp31-33) should explicitly note that the problematic elements are in red. Maybe make them bold too
Fixed.
Bugfix: Title of "Further Examples" section is surprising; "FileSweep Examples" might have been better.
Fixed.
Query: is it worth giving each example a name and not just a number?
The first paragraphs in the example sections outline the purpose of the examples. Some examples are very similar but different, so they would end up with the same crisp title unless it would be almost ridiculously long ...
Bugfix: The copyright notice should use a copyright symbol and not "(C)" where possible (really!) Here's a free one: © Don't wear it out. ;-)
Fixed.
Bugfix: The sweep namespace MUST use 'ogf' and not 'ggf'. Fix on pp4, 22-24, 26-28, 30-32, 34, 43, 45, 47. (But sweepfunc and file-sweep are both fine.)
Fixed. several occasions of "sweepfunc" were actually inconsistently using the GGF namespace. *argh* I am currently missing one final piece of information, i.e. Geoff's affiliation details to put into the final draft, before publishing draft 22, which I consider ready for submission. As soon as I have Geoffs details I will publish the draft to the Mailing list and Gridforge. Cheers, Michel -- Michel Drescher Fujitsu Laboratories of Europe, Ltd. Hayes Park Central Hayes End Road Hayes, Middlesex UB4 8FE Reg. No. 4153469 +44 20 8606 4834 Michel.Drescher@uk.fujitsu.com