39957 - 'Save as...' not fetching from cache

Status: VERIFIED FIXED

(SeaMonkey :: UI Design, defect, P1)

Description

[Paul LeBeau]

'Save as..' appears to be refetching the page rather than
retrieving it from the cache.

I'm not sure this is happening every time (I have a fast link
so it's not always obvious :), as I have had at least one
saved file which was truncated (after first fetch was
interrupted).  This tends to suggest that either the cache
*was* being used - or the truncated file size (rather
than the correct size) was being remembered.

Comment 1

[Paul LeBeau]

Forgot to mention - using M15.  Don't know about other versions.

Comment 2

[Matthew T (active 1999-2002)]

Save, Print, Send Page, etc should always use the cached version, regardless of 
the document's caching settings. Otherwise the version being saved/printed/
forwarded could be completely different from the version on the screen, which 
would be bad.

Comment 3

[WD]

Yes this is a problem.     It's one of the things that I've hated with
Netscape...    If I want to print a page that I'm looking at right now, why on
earth would it request a new copy from the network?    I want to print (or save,
forward, whatever) what I've got on the computer at that moment.   (in the cache)

There have been a couple of bugs similar to this one filed which ended up
getting marked as dupes of bug 21137.    Personally, I don't see the connection.
  I'm confirming this as new.

Comment 4

[davidm]

This isn't really a cache bug. The save as peopple and print people should be 
setting the channel flag not to do any validation which will cause the cache to 
be used if the data is available ( it could be flushed if you have multiple 
windows open). I can't actually see where they are loadin the URLs but thats the 
theory;)

Comment 5

[neeti]

Bill, 
I am not sure if you are the right person for this bug. Please reassign as 
necessary. I think we should be using SelectFileAndTransferLocation(.., instead 
of SelectFileAndTransferLocationSpec(..) in function savePage in navigator.js.

Neeti

Comment 6

[Stephen Moehle]

It would be nice to get this fixed sooner rather than later.  I cannot save any
page that was created as the result of POST data.  It seems saving the page
submits the URL but without any of the POST data so it always saves some default
page.

For instance, go to
http://www.oreillynet.com/meerkat/
and enter a number in the "Show me" field that produces more than 50 results
which would be more than one page.  Go to the second page and do "File/Save Page
As".  The first page will be saved not the second page displayed in Mozilla.

Comment 7

[Dan Rosen]

*** This bug has been marked as a duplicate of 6119 ***

Comment 8

[Jesse Ruderman]

i think this should be kept as a separate bug from bug 6119.  adding dataloss 
keyword.

should this be marked as dependent on bug 40867, like bug 6119 is?  or would it 
be sufficient to use a cached version?  (i guess this depends mostly on whether 
we always cache everything.)

Comment 9

[Matthew T (active 1999-2002)]

No, you shouldn't get a new version even if you have zero cache. Or you might 
want to save a form submission where the server told Mozilla not to cache it, for 
example.

Comment 10

[leger]

ccing gagan, updating QA contact.  Putting on [dogfood+] radar and nominating 
for nsbeta3. I use this constantly. Must work for my dogfood. :-)

Comment 11

[johng]

nav triage team:
nsbeta3+, P1

Comment 12

[Bill Law]

I've added code that does SetLoadAtributes( nsIChannel::VALIDATE_NEVER ); and it
sort of works.  I'm still seeing scenarios where either the wrong page is coming
from the cache or it's going back to the server.

For example, my fix does not fix the problem with oreillynet.com/meerkat.

Comment 13

[Bill Law]

Attachment: Fix to use VALIDATE_NEVER when saving page/image/bg-image/unk-content.

Comment 14

[Bill Law]

I'm going to assign this to Networking again.  I've fixed my code to use 
VALIDATE_NEVER whenever we reload data to save it.  As far as I know, that's the 
best I can do for now.

A problem still exists, as demonstrated by the oreillynet meerkat page.  I don't 
understand exactly what's happening.

I'm going to talk to Radha on Monday morning and try to find out what's going 
on.  It's a situation somewhat akin to session history forward/back.

Comment 15

[Dan Rosen]

adding cc

Comment 16

[leger]

PDT agrees P1

Comment 17

[Bill Law]

We have a problem resubmitting loads for any page loaded with form post data.

This applies to view-source and saving pages.

I've been talking to Radha about how to get the post data stream from session 
history by way of the docshell, but it has proven to be difficult.

Comment 18

[leger]

Putting [pdtp1] in whiteboard

Comment 19

[Bill Law]

I've checked in the patch I posted previously, plus code to resupply the form 
post data, if required.

There is still bug 20843 that can cause problems.  This produces a cache miss or 
mishit for pages involving form post data.  This seems to happen with that 
oreillynet/meerkat page (the cache erroneously returns the first page with 
matching url even though that one has different form post data).

Comment 20

[sairuh (rarely reading bugmail)]

should prolly go to tever for verification...

Comment 21

[Tom Everingham]

verified:
WinNT 1024
Linux rh6 1024
Mac os9 1024

Comment 22

[Deven Corzine]

As with bug 6119, this bug isn't really fixed.  Or, I should say, it's "fixed"
in that it's loading from the cache if possible, but not fixed in the sense of
always behaving as the user expects "Save As" to behave.  This bug should depend
on bug 40867, since it's a more fundamental problem.