239038 - rev IIDs for all modified interfaces

Status: RESOLVED FIXED

(SeaMonkey :: General, defect)

Description

[Darin Fisher]

rev IIDs for all modified interfaces.

see bug 238219 for an example of the kind of crash that can occur when we modify
a non-frozen interface without rev'ing its IID.  with all the recent
deCOMtamination work done during the 1.7 cycle, we should be sure that we have
rev'd all modified interfaces.

Comment 1

[Darin Fisher]

Some comments from jst and sicking in bug 238219:

> ------- Additional Comment #10 From Johnny Stenback  2004-03-23 18:17 PST
> I don't have a list of all interfaces that have been changed... But I bet
> nsIScriptContext and nsIScriptGlobalObject need the same treatment here. And
> maybe nsIContent too.
>
>
> ------- Additional Comment #11 From Jonas Sicking (IBM) 2004-03-23 18:36 PST
> Definitly nsIContent, nsIHTMLContent and nsIStyledContent too. Maybe 
> nsINodeInfo and nsINodeInfoManager too.

jst: if you don't have time to own this, then bounce it back to me.  i just
figured that you probably had your finger on the pulse of most of the
deCOMtamination work.

Comment 2

[Benjamin Smedberg]

a bonsai query since 11-Dec-2004 (1.6 branch date) indicates the following
interfaces have been changed without an IID rev:

browser/components/bookmarks/public/nsIBookmarksService.idl
mailnews/compose/public/nsIMsgAttachment.idl
mailnews/compose/public/nsIMsgCompose.idl
xpfe/components/download-manager/public/nsIDownloadManager.idl
mailnews/base/public/nsISubscribableServer.idl
xpcom/obsolete/nsIFileSpec.idl (methods added at end of interface only)
dom/public/idl/html/nsIDOMHTMLButtonElement.idl
dom/public/idl/html/nsIDOMNSHTMLButtonElement.idl
mailnews/db/msgdb/public/nsIMsgDatabase.idl
mailnews/base/public/nsIMessenger.idl
editor/composer/public/nsIEditingSession.idl
xpcom/threads/nsIProcess.idl
extensions/cookie/nsICookiePromptService.idl
dom/public/idl/css/nsIDOMCSS2Properties.idl
gfx/idl/nsIFreeType2.idl
dom/public/idl/events/nsIDOMPopupBlockedEvent.idl
dom/public/idl/base/nsIDOMChromeWindow.idl
docshell/base/nsIContentViewer.idl
netwerk/base/public/nsIPasswordManagerInternal.idl
dom/public/idl/base/nsIDOMCrypto.idl
extensions/xmlextras/base/public/nsIXMLHttpRequest.idl
extensions/xmlextras/base/public/nsIDOMSerializer.idl

I only did IDL, not .h pseudo-interfaces, but these pseudo-interfaces should
also probably be revved:
nsICSSLoader
nsICSSStyleRule
nsIScriptContext
nsIHTMLDocument
nsIParserService
nsIDOMLoadListener

Comment 3

[Darin Fisher]

-> me

Comment 4

[Darin Fisher]

Attachment: v1 patch

This patch changes the IIDs for all interfaces and pseudo-interfaces in
bsmedberg's list.  I did not change the IID of nsIFileSpec since I do not think
there is any chance that external code is implementing that interface.	I also
did not change the IID for the frozen nsIDOMHTMLButtonElement since the change
to that interface did not change the ABI.  It only marked a method [noscript]
that was previously scriptable.

Comment 5

[Johnny Stenback (:jst)]

Comment on attachment 146212 [details] [diff] [review]
v1 patch

sr=jst. Thanks for doing this, Darin!

Comment 6

[Darin Fisher]

fixed-on-trunk

Comment 7

[Brendan Eich [:brendan]]

Comment on attachment 146212 [details] [diff] [review]
v1 patch

a=[e-mail] for 1.7 final.

/be

Comment 8

[Darin Fisher]

fixed1.7

Comment 9

[Michiel van Leeuwen (email: mvl+moz@)]

After this change,
http://lxr.mozilla.org/seamonkey/source/extensions/cookie/nsCookiePromptService.h#56
doesn't match with
http://lxr.mozilla.org/seamonkey/source/extensions/cookie/nsICookiePromptService.idl#46
anymore. (#define NS_COOKIEPROMPTSERVICE_CID is not updated)

Shouldn't they match? Now, nsCookiePromtService implements the wrong CID. Or am
I confused about xpcom? (not an unlikely possibility...)

I only looked at this interface, because i remembered that interface having a
#define in the .h file. Should all the changes be checked for this?

Comment 10

[Christian :Biesinger (don't email me, ping me on IRC)]

(In reply to comment #9)
http://lxr.mozilla.org/seamonkey/source/extensions/cookie/nsCookiePromptService.h#56
> doesn't match with
>
http://lxr.mozilla.org/seamonkey/source/extensions/cookie/nsICookiePromptService.idl#46
> anymore. (#define NS_COOKIEPROMPTSERVICE_CID is not updated)

CIDs and IIDs don't have any relation, are you saying that they were the same
before? that seems wrong.

> I only looked at this interface, because i remembered that interface having a
> #define in the .h file. Should all the changes be checked for this?

It did? The link you posted shows that only the component has?

Comment 11

[Michiel van Leeuwen (email: mvl+moz@)]

Ok, so i was confused. Ignore my comment :)
On the other hand, i always though iid = cid for some reason. So i need to check
if i made them the same in other places too.....

Comment 12

[Darin Fisher]

(In reply to comment #11)
> Ok, so i was confused. Ignore my comment :)
> On the other hand, i always though iid = cid for some reason. So i need to check
> if i made them the same in other places too.....

It's not a big deal if IIDs and CIDs conflict.  They are never used in the same
context.