74738 - [FIX] not printing out full directions

Status: VERIFIED FIXED

(Core :: Printing: Output, defect)

Description

[sujay]

using 4/4 build of netscape

1) launch netscape
2) jump to Netscape home page
3) click on Maps
4) enter FROM and TO fields
5) Click Map It!
6) File | Print

go look at the output. you'll notice it prints out only partial
driving directions.

Comment 1

[Kevin McCluskey (gone)]

Sujay: Is this happening with a recent build?

Comment 2

[sujay]

I cannot verify this because of this blocker bug 77733

Comment 3

[dcone (gone)]

fixed with the blank page problem.

Comment 4

[sujay]

still not working...I only get half of the driving directions...

tried on 5/9 build on windows.

Comment 5

[dcone (gone)]

this is a table pagination bug.. its breaking the pages up wrong..

Comment 6

[karnaze (gone)]

I got directions from 12233 Pomerado Rd, San Diego, CA to 4000 El Camino Way, 
Palo Alto, CA and not all of the directions printed out and the map was started 
on the 2nd page, wasting a lot of paper. Adding dataloss keyword.

Comment 7

[karnaze (gone)]

Attachment: test case where align=left table does not print fully

Comment 8

[karnaze (gone)]

Reassigning to attinasi and moving to m0.9.2 (although it may not make it until 
m0.9.3). It appears that blocks always reflow a floater with an unconstrained 
height during printing, which doesn't allow the floater to return an incomplete 
status. I tried a simple fix in nsBlockFrame::ReflowFloater, substituting 
NS_UNCONSTRAINEDSIZE with aState.mAvailSpaceRect.height, but it didn't work.

Comment 9

[Chris Waterson]

Is this a regression from my floater changes? If so, I'd be happy to look at 
it...

Comment 10

[karnaze (gone)]

Attachment: simpler test case

Comment 11

[Marc Attinasi]

The testcase was broken back to NS6.01, although the driving instructions
crashed NS6.01 so I couldn;t verify that it was broken back then. It is not a
regression from waterson's changes though, it is just plain broken.

Accepting and investigating now.

Comment 12

[Marc Attinasi]

karnaze, I noticed that with the testcase you attached there is a problem even
without the floating table (no align=left attr). The second page comes out
blank. Of course, with the table floating there is no second page at all.

Comment 13

[karnaze (gone)]

Marc, it looks like a table with a height (maybe the height needs to be larger 
than the natural height of the table) doesn't split correctly. I filed bug 
85738. I also verified that the original page of driving directions does not 
have a table with a specified height, so bug 85738 should not block you. 

Comment 14

[karnaze (gone)]

So, the 1st attachment is a valid test case that doesn't hit bug 85738.

Comment 15

[Marc Attinasi]

Bug 85768 has been opened to on the larger issue: any floated block is not
continued on the next page. The table with the driving instructions is a special
case of that problem.

Comment 16

[Marc Attinasi]

I have a temporary fix for this bug. Until the larger problem of floated blocks
not continuing is solved, I put in a '@media print' rule to un-float aligned
tables. The testcase, the driving instructions, and many other sites using long
tables now print fine.

I'll attach a patch for comments and/or ridicule.

Comment 17

[Marc Attinasi]

Attachment: Temporary hack to html.css to unfloat floated tables when we print so that we get the whole table

Comment 18

[Marc Attinasi]

status whiteboard graffiti: need reviews for temporary patch, please.

Comment 19

[Chris Waterson]

Heh. Okay with me. [r/sr]=waterson

Comment 20

[Kevin McCluskey (gone)]

r=[e-mail]

Comment 21

[Christopher Blizzard (:blizzard)]

a=blizzard on behalf of drivers for 0.9.2.

Please make sure that in bug 85768 that you note that you need to remove the
hacks to html.css.

Comment 22

[Marc Attinasi]

OK, hack noted in depends on bug 85768

Comment 23

[Marc Attinasi]

/cvsroot/mozilla/layout/html/document/src/html.css,v  <--  html.css
new revision: 3.130; previous revision: 3.129
done

Comment 24

[sujay]

I'm verifying this bug...everything but the last page gets printed out.

should I reopen this bug ?

Comment 25

[Marc Attinasi]

Hard to say if this one should be reopened. Since removing my patch makes
MapQuest complete the printjob (even though the directions are truncated to
whatever fits on the first page), it appears that the problem is somehow related
to my patch.

I'll open a new bug on the new problem. This bug is really about continuing
tables past the first page, and that has been hacked, er, fixed as demonstrated
by the testcase (and MapQuest to some extent).

Comment 26

[sujay]

Thanks Marc, can you Cc: me on the new bug?

I will close this out and mark it VERIFIED-FIXED....

Comment 27

[sujay]

Marc opened a new bug for the other issue, bug 87374

marking this one verified-fixed.