74726 - [API] Make |nsSharable[C]String| writable and add xpidl-string functionality

Status: RESOLVED WONTFIX

(Core :: XPCOM, defect)

Description

[Scott Collins]

See comments in bug #53065.  Additionally, we'll need to deploy this class.  See
bug #70143.  This class will be the preferred string class for general
operations, and for sharing across JS boundaries.

Comment 1

[Scott Collins]

Also, rename to |nsSharable[C]String|.

Comment 2

[Scott Collins]

All planned string API fixes need to be in by mozilla1.0.

Comment 3

[Scott Collins]

Attachment: step one is to add some ability to adopt a new buffer, vidure needs this asap

Comment 4

[Scott Collins]

s/vidure/vidur/, sorry.  All: see bug 18843 for why I needed to add this
functionality.

Comment 5

[vidur (gone)]

r=vidur on the patch. Presumably you'll clean up later and change all callers of
|NS_AllocateContiguousHandleWithData| to use its new form?

Comment 6

[Scott Collins]

Attachment: as before, but also make appropriate constructors |explicit|

Comment 7

[Johnny Stenback (:jst)]

sr=jst

Comment 8

[Scott Collins]

this partial fix is checked in, more to come

Comment 9

[Scott Collins]

See also (from bug 75220) the patch

  http://bugzilla.mozilla.org/showattachment.cgi?attach_id=34639

...that I must checkin before I can continue in this bug.

Comment 10

[Scott Collins]

This class has been renamed, changing the summary accordingly.

Comment 11

[Scott Collins]

...and as the renaming is now accomplished, no longer depends on bug 75220.

Comment 12

[Scott Collins]

Attachment: scratch C++ showing one way to add |getter_Copies| into a |nsSharableC?String|

Comment 13

[Scott Collins]

Attachment: replaces the old implementation with a new one based on |nsSharableC?String|

Comment 14

[Scott Collins]

still want to add comments describing the mechanism and alternatives; and some
debug machinery to count instances, find high-water marks, etc.  looking for
problems in this patch where I can accidentally end up with NULL

Comment 15

[Scott Collins]

Attachment: (replaces patch above) adds debug analysis machinery and comments describing the underlying mechanism and alternatives

Comment 16

[Scott Collins]

Attachment: (replaces all patches above) fixed obvious |AddressOfDataStart| bug, eliminated assertions for |NULL| handles

Comment 17

[Scott Collins]

This code is passing all my tests.  I believe the patch above is ready for review.

Comment 18

[Chris Waterson]

r=waterson

Comment 19

[Patrick C. Beard]

r=beard

Comment 20

[Scott Collins]

Attachment: (replaces all patches above) adds additional debug-statistics machinery to count cases where sharing is attempted

Comment 21

[Scott Collins]

Attachment: (replaces all patches above) more debug-statistics machinery

Comment 22

[Scott Collins]

I'm going to add all the assignment operators etc., needed to make both these
classes |nsSharableC?String| and |nsXPIDLC?String| writable (as per the bug
summary) before checking in.  I'll get this re-reviewed after attaching those
patches.

Comment 23

[Scott Collins]

Attachment: (replaces all patches above) ensures that string preprocessor symbols are always defined (one way or the other), changes |#ifdef|s to |#if|s, and adds appropriate flags to MozillaBuildFlags.txt for Mac (other platforms set on the command line)

Comment 24

[Scott Collins]

On second thought, I'm just going to check this in as soon as the tree is open.
 The assignment stuff is almost ready to go, but I don't want to stop this patch
for that.  I'll continue to use this bug for the patches.

Comment 25

[Scott Collins]

Attachment: (replaces all patches above) factors the |#define|s, and ensures they are seen by all string files

Comment 26

[Scott Collins]

Attachment: possible work-around for |nsCStringKey| problems

Comment 27

[Jean-Francois Ducarroz]

cc'ing myself as I need to be able to append nsXPIDLStrings.

Comment 28

[David Baron :dbaron: (⌚️UTC-4, no longer working on Mozilla)]

Bug 104663 has been checked in, so nsSharableC?String should now be writable.

Comment 29

[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 30

[Scott Collins]

giving up ancient string bugs to the new string owner.  jag, you'll want to sort
through these and see which ones still apply and go with or against the
direction in which you intend strings evolve

Comment 31

[Vedran Miletic]

retargeting

Comment 32

[jag (Peter Annema)]

This work is no longer relevant, strings have had a new implementation for a while now :-)