122078 - Save as complete page sometimes stalls while saving framesets

Status: VERIFIED FIXED

(Core Graveyard :: File Handling, defect)

Description

[Dan Tobias]

While trying to create a simple test case for bug 122077, I ran into this
unrelated problem.  When I try to use "Save Page" to save the complete page of
this simple frameset, the action stalls with the "Saving" dialog showing.  When
I finally cancel the dialog, I see that the subdirectory structure to handle
saving the different frames has been created, but no files have been saved in
it.  (This is different from the dialog hanging around on saving as plain text,
since in that case when you dismiss the dialog the saving has in fact completed.)

Comment 1

[Boris Zbarsky [:bzbarsky]]

Dan, what build are you using?  A similar bug was recently fixed...

Comment 2

[Dan Tobias]

Currently, build 2002012403... but I update regularly, so I'll try a new build.

Comment 3

[Boris Zbarsky [:bzbarsky]]

No, that's recent enough.  :)

Over to adamlock.

Comment 4

[Adam Lock]

Attachment: Patch

Reviews please?

Patch is quite simple. I've added some code to count the number of persistable
entries in the mURIMap rather than use the total so that the "non URIs to save"
path is followed properly.

Comment 5

[Adam Lock]

Bill or Boris, can you review this patch please? Thanks.

Comment 6

[Kathleen :Brade]

Comment on attachment 66754 [details] [diff] [review]
Patch

r=brade

Comment 7

[Bill Law]

Seems like "urisToPersist" could be a data member, then EnumCountURIsToPersist
could be merged with EnumPersistURIs and we'd only have to make one pass over
the list.

That would be of some negligible performance benefit, but I'm not sure it's a
better or worse approach; just throwing it out there.

Comment 8

[Adam Lock]

I could probably have done it in EnumPersistURIs, but I really need to know
up-front whether there are URIs to persist so I know which path to take - state
notifications and document saving done asynchronously via stream listeners, or
directly. So I think it's clearer to do it this way a take a small performance hit.

Comment 9

[rpotts (gone)]

Comment on attachment 66754 [details] [diff] [review]
Patch

sr=[e-mail]

Comment 10

[Adam Lock]

Fix is in

Comment 11

[sairuh (rarely reading bugmail)]

vrfy fixed using 2002.03.13 comm bits on win2k, mac 10.1.3 and linux rh7.2.
thanks for the test case, Dan!