78013 - eliminate nsFileSpec usage from nsRDFXMLDataSource

Status: RESOLVED FIXED

(Core Graveyard :: RDF, defect, P4)

Description

[Chris Waterson]

I did this a while ago. The diffs are rotting in my tree.

Comment 1

[Chris Waterson]

Attachment: eliminate nsFileSpec usage, rename mURLSpec to mOriginalURLSpec

Comment 2

[Chris Waterson]

The above patch:

- renames mURLSpec to mOriginalURLSpec, to make its purpose a bit more clear

- Uses nsIFile and friends instead of nsFileSpec to write the RDF/XML back
  to disk.

dougt or darin, could you r=? shaver, sr=?

Comment 3

[Doug Turner (:dougt)]

r=dougt

Comment 4

[Mike Shaver (:shaver emeritus)]

+    // Is it a file? If so, we can write to it. Some day, it'd be nice
+    // if we didn't care what kind of stream this was...

Hasn't that day come?  You can write to an FTP URL and it'll upload, at least. 
Dougt? Darin?

Can mOriginalURLSpec be an nsXPIDLCString?

Comment 5

[Doug Turner (:dougt)]

shaver, that day has not come yet.  It really has not been a high priority :-(

Comment 6

[Chris Waterson]

Attachment: eliminate nsFileSpec, change mOriginalURLSpec to an nsCString

Comment 7

[Chris Waterson]

shaver: what dougt said. I talked at length with darin about this; his belief is
that there needs to be a layer above the channel that would provide a ``virtual
file system'', and normalize read/write operations across protocols.

This code predated nsXPIDLCString, and should probably be ready for the bright
day when scc eliminates nsXPIDLCString and its ilk. I changed mOriginalURLSpec
from a bare |char*| to an nsCString.

Comment 8

[Chris Waterson]

fix checked in

Comment 9

[Chris McAfee]

reopened, this introduced bug 78718.  I've backed this patch out.

Comment 10

[Chris Waterson]

Ok, I'll try again later!

Comment 11

[Chase Tingley]

bug 36972 may be a very old dup (vice-versa, rather), but since there appear to
be a couple instances of nsIFileSpec in /rdf/ not covered by the patch (eg
nsFileSystemDataSource), marking a dependency instead.

Comment 12

[Chase Tingley]

Most of the string changes in the original patch went in for a different bug, so
for fun I boiled down the last patch into just the nsFileSpec change and tried
it out.

After applying it, I can't reproduce any of the symptoms described in bug 78718.
 Very strange.

Comment 13

[Chase Tingley]

Attachment: boiled-down version of waterson's patch I've been testing with

Comment 14

[Asa Dotzler [:asa]]

Bugs targeted at mozilla1.0 without the mozilla1.0 keyword moved to mozilla1.0.1 
(you can query for this string to delete spam or retrieve the list of bugs I've 
moved)

Comment 15

[Robert John Churchill]

Comment on attachment 51711 [details] [diff] [review]
boiled-down version of waterson's patch I've been testing with

r=rjc

Comment 16

[Chris Waterson]

Comment on attachment 51711 [details] [diff] [review]
boiled-down version of waterson's patch I've been testing with

sr=waterson

Comment 17

[Darin Fisher]

you should really consider using a buffered output stream overtop the file
output stream.  see NS_NewBufferedOutputStream in nsIFileStreams.idl

Comment 18

[Robert John Churchill]

I checked in tingley's patch as it was needed for other bugs (113894, for 
example).  Marking bug fixed.

Comment 19

[benc]

tever is not RDF QA anymore