672055 - Use nsCharSeparatedTokenizer to parse number-optional-number attributes

Status: RESOLVED FIXED

(Core :: SVG, defect)

Comment 1

[Robert Longson [:longsonr]]

Attachment: patch

Comment 2

[Daniel Holbert [:dholbert]]

Comment on attachment 546333 [details] [diff] [review]
patch

Looks good!  One thing that I stumbled over at first, though: so it looks like lastTokenEndedWithWhitespace() would end up returning true for the first token in all the following strings:
>  "first, last"      <--- discussed below
>  "first  ,last"
>  "first  , last"
>  "first last"
(but would return false for "first,last" or "first," or "first")

However, in the initial line above, it's not intuitively clear that the first token actually "ends with whitespace" -- we're tokenizing at the ',' character, so my first impression is that the first token ends _at the comma_ (and hence does *not* "end with whitespace").

I'm not sure this distinction matters that much, though -- it's just non-obvious from the function-name that it'll return true in this sort of situation.  Could you add a header comment saying something like "Returns true if there was whitespace after the most recently parsed token. This includes whitespace after any separator character that was encountered."

r=dholbert with that clarification.  You should get someone else (maybe roc?) to sign off on the /xpcom/ds/ chunk, too.

Comment 3

[Robert Longson [:longsonr]]

(In reply to comment #2)
> Comment on attachment 546333 [details] [diff] [review] [review]
> patch
> 
> Looks good!  One thing that I stumbled over at first, though: so it looks
> like lastTokenEndedWithWhitespace() would end up returning true for the
> first token in all the following strings:
> >  "first, last"      <--- discussed below
> >  "first  ,last"
> >  "first  , last"
> >  "first last"
> (but would return false for "first,last" or "first," or "first")

I'm not sure it makes sense to call that method then. I'll try adding an NS_ABORT_IF_FALSE that we're at the end.

Comment 4

[Robert Longson [:longsonr]]

(In reply to comment #2)
> Comment on attachment 546333 [details] [diff] [review] [review]
> patch
> 
> However, in the initial line above, it's not intuitively clear that the
> first token actually "ends with whitespace" -- we're tokenizing at the ','
> character, so my first impression is that the first token ends _at the
> comma_ (and hence does *not* "end with whitespace").

Actually I can have the flag return fals in that case as I don't really care what the value is if lasttokenendedswithseparator is true.

> 
> I'm not sure this distinction matters that much, though -- it's just
> non-obvious from the function-name that it'll return true in this sort of
> situation.  Could you add a header comment saying something like "Returns
> true if there was whitespace after the most recently parsed token. This
> includes whitespace after any separator character that was encountered."

I'll fix the logic instead so it's more intuitive and get roc to sr.

Comment 5

[Robert Longson [:longsonr]]

Attachment: address review comment

Comment 6

[Robert Longson [:longsonr]]

Attachment: address review comment properly

Comment 7

[Marco Bonardo [:mak]]

http://hg.mozilla.org/mozilla-central/rev/7a9c173b0898