[wp-trac] [WordPress Trac] #16746: post paginating <!--nextpage--> does not work

[WordPress Trac]

#16746: post paginating <!--nextpage--> does not work
--------------------------+-----------------------
 Reporter:  hakre         |       Owner:  nacin
     Type:  defect (bug)  |      Status:  reopened
 Priority:  normal        |   Milestone:  3.6
Component:  Query         |     Version:
 Severity:  normal        |  Resolution:
 Keywords:  needs-patch   |
--------------------------+-----------------------
Changes (by nacin):

 * keywords:  has-patch commit => needs-patch


Comment:

 I did understand the initial report here, and based on my original
 understanding of the surrounding code, I did not think this patch should
 be applied. Basically, if someone had `<!--nextpage-->` at the start of a
 post, and then later in the post, they'd currently have two pages. With
 this patch, they would have three — including an empty first page. This is
 not expected and is fairly lame. While a "page break" usually means to
 immediately start a new page, `<!--nextpage-->` just as much marks the
 start of a page, and if a user was using it to mark the start of page 1
 (and it worked) then they should not be punished. I also assumed, knowing
 the age of this code, that this check existed since b2.

 However, in this case, I was wrong. While this code is from b2,
 [http://core.trac.wordpress.org/browser/trunk/b2-include/b2template.functions.php?rev=20#L1237
 b2 used a preg_match] that *did* obey `<!--nextpage-->` at the start of a
 string. It was changed to strpos() in [8754], for performance but it also
 changed the functionality.

 Additionally, this is check is to decide whether there is pagination at
 all. If the post starts with `<!--nextpage-->`, strpos() will return 0,
 and no other `<!--nextpage-->` will be listened to, and the post will not
 be paginated. That is clearly incorrect behavior.

 While failing to explicitly and strictly check the return value of
 strpos() is a code smell, it helps when bug reports point out ''why''
 something smells, as well as any repercussions. In this case, I had felt
 that the implicit > 0 (needle in haystack, and needle not at start of
 haystack) was correct.

 So — I think this would be best with pieces of both the current and old
 behaviors: Ignore `<!--nextpage-->` at the start of the string (so, no
 empty page), but don't let it prevent pagination.

-- 
Ticket URL: <http://core.trac.wordpress.org/ticket/16746#comment:8>
WordPress Trac <http://core.trac.wordpress.org/>
WordPress blogging software