134066 - nsIPromptService does not support accesskey

Status: VERIFIED FIXED

(Core :: DOM: UI Events & Focus Handling, defect)

Description

[Aaron Leventhal]

When using nsIPromptService::ConfirmEx() for example, there is no way to ask for
buttons that have underlined letters (accesskey's). Thus there is no way to
associate Alt+S with _S_ave, for example.

Comment 1

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

I'm pretty sure that the cookie accept dialog uses nsIPromptService as well, so
marking this as blocking bug 107560

Comment 2

[Aaron Leventhal]

-> ccarlen

Comment 3

[Aaron Leventhal]

Suppose XUL accesskeys could be coded using an entity from a DTD, like &_; right
before the letter to be underlined..
For example:
Save &_;As

To make it legal, there would have to be a DTD entry:
<!ENTITY _ "">

Then, wherever it is that we decode entities, we would SetAttribute(accesskey,
letter) when we saw thsi entity, in order to implement it.

This would make some things easier, make it possible to remove some code, and
allow nsIPromptService to support accesskey without adding additional arguments.

Is this a good idea?

Comment 4

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

why not make it much simpler and use, say, only an underscore like:
Save _As

Comment 5

[Aaron Leventhal]

Because presumeably _ is a legal character, that someone might want in their text.

Comment 6

[Aaron Leventhal]

Just spoke with danm about it. 

nsIPromptService isn't frozen. Perhaps we can do something to it without messing
up to many people.

If we choose & (as Windows does), then we'll probably need to make \& work if
they really do want an &.

Dan, would this affect a lot of embeddors?

Comment 7

[Akkana Peck]

*** Bug 64457 has been marked as a duplicate of this bug. ***

Comment 8

[Akkana Peck]

*** Bug 61877 has been marked as a duplicate of this bug. ***

Comment 9

[Charles Manske]

Is this work scheduled? Can we get a milestone on this?

Comment 10

[Aaron Leventhal]

I told ccarlen I would do it. I know, this is important work -- until another
engineer can help out on accessibility/keyboard issues, I can't say when it will
get done. Or, ccarlen, maybe you have time for it?

Comment 11

[Aaron Leventhal]

Attachment: Allows users of nsIPromptService to insert "&" before a char to make it the accesskey (&& escapes if you really want &)

Also fixes bug 133582, so that simple dialogs with "Don't ask me again"
checkboxes don't start with initial focus on checkbox, but on first button
instead.

Seeking r=

Comment 12

[Samir Gehani]

Comment on attachment 101196 [details] [diff] [review]
Allows users of nsIPromptService to insert "&" before a char to make it the accesskey (&& escapes if you really want &)

r=sgehani

Comment 13

[Joe Hewitt (gone)]

Comment on attachment 101196 [details] [diff] [review]
Allows users of nsIPromptService to insert "&" before a char to make it the accesskey (&& escapes if you really want &)

sr=hewitt

Comment 14

[Aaron Leventhal]

taking

Comment 15

[Aaron Leventhal]

checked in

Comment 16

[[e-mail]]

Comment on attachment 101196 [details] [diff] [review]
Allows users of nsIPromptService to insert "&" before a char to make it the accesskey (&& escapes if you really want &)

>+  var accessKeyIndex;
>+  if (aLabel.charAt(0) == '&') {
>+   accessKeyIndex = 0;
>+  }
>+  else {
>+    accessKeyIndex = aLabel.search(/[^\&]\&[^\&]/) + 1;
>+    if (accessKeyIndex == 0) {
>+      accessKeyIndex = -1; // magic value for no accesskey
>+    }
>+  }
Edge case: "&&" becomes "&" with accesskey="&"?

Comment 17

[Aaron Leventhal]

Neil, if you mean that if the string starts with &&, you're right -- I see now
that it would make an accesskey of &.

However, if && is in the middle it works to create a real &

Comment 18

[sairuh (rarely reading bugmail)]

rs vrfy.

Comment 19

[Asa Dotzler [:asa]]

*** Bug 186443 has been marked as a duplicate of this bug. ***