(No text was entered with this change)

To make copying work, we could include the original text as a hidden element near the
ellipsis, and make the ellipsis uncopyable (user-select: none, or make it "content"
in a :before/:after rule).
This could hypothetically also be used to fix issue 5928 (haven't tested).

I tested the thing in comment 3, and it more-or-less works. Markup/CSS:
<span class="value">url("abcde<span class="ellipsis">fghijklmnopqrstu</span>vwxyz")</span>
.value > .ellipsis { font-size: 0; }
.value > .ellipsis::after { content: '...'; font-size: 13px; }

Then:
The "..." is unselectable.
$(".value").textContent == 'url("abcdefghijklmnopqrstuvwxyz")', and copy-paste preserves
that.
When hovering 'url("abcde': rangeOffset ranges from 0 to 9, rangeParent = text-node
of 'url("abcde'
When hovering '...': rangeOffset is 16 (length of rangeParent.textContent), rangeParent
= text-node of 'fghijklmnopqrstu'
When hovering 'vwxyz")': rangeOffset ranges from 0 to 6, rangeParent = text-node of
'vwxyz")'

So that seems to work rather nicely, though it complicates the domplate and the rangeOffset
calculations a bit.

1) Related pull request:
https://github.com/firebug/firebug/pull/49

2) Related test failures:
css/5277/issue5277.html
css/5461/issue5461.html
(as mentioned here http://code.google.com/p/fbug/issues/detail?id=5862#c18)

---

Fix for this issue should be in Firebug 1.11a5

Honza

> So that seems to work rather nicely, though it complicates the domplate and the 
> rangeOffset calculations a bit.
So or so copying will be a bit more complicated.
The only other solution I can imagine is to catch the copy commmand and replace it
by code that fetches the data behind it. This method would actually be better, because
it would allow to control the formatting of the output (having Fireformat in mind).
Non-the-less it's a nice approach. So Farshid, please take Simon's code for fixing
the copy & paste problem into consideration. When implementing that, please add some
comments describing this workaround.

Sebastian

> The only other solution I can imagine is to catch the copy commmand and replace it
> by code that fetches the data behind it.
... which is made harder by the fact that arbitrary text parts can be selected (e.g.,
a whole rule, multiple rules, a property, a property value, or part of a property value),
and that Ctrl/Command+C is remappable.

> ... which is made harder by the fact that arbitrary text parts can be selected
I didn't say the solution would be easier. :-) Of course it can get quite complex.
That's why I suggested we implement your solution. We could still advance it in a later
step if we wish to do so.

Sebastian

<span class="value">url("abcde<span
> class="ellipsis">fghijklmnopqrstu</span>vwxyz")</span>
> .value > .ellipsis { font-size: 0; }
> .value > .ellipsis::after { content: '...'; font-size: 13px; }
>
> Then:
> The "..." is unselectable.
> $(".value").textContent == 'url("abcdefghijklmnopqrstuvwxyz")', and
> copy-paste preserves that.
>
Simon, Could you explain more, please? How is this possible while you used
another span with class 'ellipsis' above , but you assign the
URL value to the textContent directly at this line of code.

Farshid

> but you assign the URL value to the textContent directly at this line of code.
I'm not assigning, just comparing. What I mean is that the expression
$(".value").textContent == 'url("abcdefghijklmnopqrstuvwxyz")'
returns true with the above HTML.

.ellipsis spans of course needs to be created manually.

> I'm not assigning, just comparing.
Ah, mistake by me.

> What I mean is that the expression
> $(".value").textContent == 'url("abcdefghijklmnopqrstuvwxyz")'
> returns true with the above HTML.
I see.

> .ellipsis spans of course needs to be created manually.

I created another template in CSSPropTag, for creating ellipsis span and use in tag
template (to embed in it as value) but i couldn't do with replace() and append() functions,
because i didn't use domplate in this way. Simon, Could you please give me an example,
if you have any idea? thanks!

Farshid

I don't know domplate very well, sorry. Honza should know.

Sure, I can help with Domplate what exactly is the problem?

Honza

Honza, I need a template to implement what Simon wrote at comment #4
http://code.google.com/p/fbug/issues/detail?id=5880#c4
A template for splitting cropped string into two parts and a span keeping removed string
of cropped string between them . Another problem is that this template should work
with both cropped values and non-cropped values(because the URL long values only should
be cropped). Thanks!


Farshid

> A template for splitting cropped string into two parts and a span keeping removed

> string of cropped string between them .
Note that Str.cropString() accepts a third parameter defining the alternative text,
which could be set to "<span class=\"alterText\">...</span>" in our case. The hard
part is to integrate that into the Domplate. I tried using Dom.appendInnerHTML() for
that purpose but failed so far.

Sebastian

> Note that Str.cropString() accepts a third parameter defining the alternative text,
> which could be set to "<span class=\"alterText\">...</span>" in our case.
Well, it's not that simple. We only want to crop url() values here, and we don't want
to replace cropped parts with the fixed string "<span class=\"alterText\">...</span>"
but with "<span class=\"alterText\">" + croppedValue + "</span>". But I agree that
this isn't the hard part.

> Note that Str.cropString() accepts a third parameter defining the alternative text,
> which could be set to "<span class=\"alterText\">...</span>" in our case. The hard
> part is to integrate that into the Domplate.
That is why i said we need a template for splitting the cropped string into two parts
and a span element keeping removed string of cropped string between them, because the
value of the SPAN() function is behaved as pure text, so i think we have to create
.ellipsis span with SPAN() function of the domplate and two other SPAN() function for
the first half and the last half of the cropped string.
As i said before, the Problem is, this template doesn't work when There may be other
things besides a URL in a property value or there isn't a URL value at all.

I tried several ways, but i couldn't find the solution yet. Honza, could you suggest
something?

Farshid

So, first I couldn't pass STR in comment #0

> The new property will be merged with the "background" shorthand property
> and cropped to something like "url(...een". Hovering "een" doesn't show
> the infotip for the color.
When I create 'background-color' prop, the original 'background' disappears.

I tested with firebug/master and FarshidB/master

> Honza, could you suggest something?
I don't see any magic solution for this. My recommendation would be to use Domplate
to generate just the basic structure, i.e. keep SPAN({"class": "cssPropValue"}) as
it is and dynamically (manually) create the inner SPANs (mimicking what Simon suggested
in comment #4).

This should be done at the right moment, keeping in mind that CSSPropTag.tag is also
reused in other templates.

Honza

> When I create 'background-color' prop, the original 'background' disappears.
Issue 6016 - I'll fix it. In the mean time, try typing faster. ;)

>> Honza, could you suggest something?
> I don't see any magic solution for this. My recommendation would be to use Domplate

> to generate just the basic structure
FYI my solution of comment 16 is currently just failing because it requires Domplate
to allow passing multiple parameters to a function:

http://www.christophdorn.com/Blog/2008/09/10/domplate-variables-and-variable-formatters/

Unfortunately Christoph never applied his changes to the Firebug Domplate and I can't
seem to get it to work.

Sebastian

> FYI my solution of comment 16 is currently just failing because it
> requires Domplate to allow passing multiple parameters to a function
This sounds promising!

I just committed a patch for issue 5107, it works in my simple test case:

// Domplate
var firstTemplate = domplate(
{
    tag:
        DIV({"class": "greenLabel"}, "$label1,$label2|getLabel"),

    getLabel: function(label1, label2)
    {
        return label1 + ", " + label2;
    },
});

// Template execution.
var label1 = "Hello World 1!";
var label2 = "Hello World 2!";

var parentNode = $("#content")[0];
firstTemplate.tag.replace({label1:label1, label2:label2},
    parentNode, firstTemplate);


Does it work in your template?

Honza

> Does it work in your template?
Yes, it works. Thanks a lot, Honza.

Farshid asked me in an email:
> i saw that you commented on issue that if Domplate supports
> passing more than a parameter to the functions, the problem would be solved.
> Could you explain how it could
Basically it's what I described in comment 16. I attached a work-in-progress patch
based on Farshid's github repo.

Also I want to note that copying is not the only issue we have with this. Also the
search won't work anymore on these long URLs.

Sebastian

> return fragment;
I don't understand how this works. Domplate magic?

'value' in getCroppedUrlValue needs HTML escaping (but note the pitfall of escaping
", which could make url() parsing weird).

> + return match.replace(p1, Str.cropString(p1, length, "<span class=\"alterText\">...</span>"));
As mentioned in comment 17, this is wrong; "..." needs to be the actual text for the
hack to function. Better do cropping manually here. (Could also avoid breaking escaped
HTML into parts, and make the replacing more efficient.)

And of course if we go this route, showInfoTip needs fixes.

> Also the search won't work anymore on these long URLs.
Ah, true! Does this method fix that?

> I attached a work-in-progress patch based on Farshid's github repo.
Sebastian, I couldn't apply the patch. After trying to apply the patch, github UI says
'fatal: corrupt patch at line 255', so i opened that with a text editor and saw the
code. I noticed you use createContextualFragment() to create a fragment and return
the fragment at the end of the function, but i tried the same way a few days ago with
createDocumentFragment() function and it doesn't work (there was text '[Object DocumentFragment]'
instead of actual value), so before trying I'd like to ask you that are you sure it
works? you tested it?
Could you answer me before meeting, please? Because Honza sent me an email today and
wrote in that we want to discus this on today's call, but if this is solved, so we
need so and I can get on with this issue based your changes. Thanks!

Farshid

>> return fragment;
> I don't understand how this works. Domplate magic?
>
> (there was text '[Object DocumentFragment]' instead of actual value), so before 
> trying I'd like to ask you that are you sure it works
I wrote that it's a WIP, so no, it's not working yet.

> As mentioned in comment 17, this is wrong; "..." needs to be the actual text for
the hack to function.
Ah, right.

>> Also the search won't work anymore on these long URLs.
> Ah, true! Does this method fix that?
No, that would need to be fixed in a separate issue.

It seems you were discussing this extensively in the conf call. I agree with reverting
all changes made so far regarding value cropping and put the changes into a new branch.
We need to rethink the whole concept of this. We need to find a simple, reusable solution,
so we can also fix things like issue 1469.

Sebastian

> No, that would need to be fixed in a separate issue.
Hm, are you sure? Honza mentioned that we'd probably also want "expansion" functionality
when finding matches within cropped strings, so that's missing, but it looked like
searching uses Firefox's built-in search function, and that works (modulo https://bugzilla.mozilla.org/show_bug.cgi?id=804799
).

> [Object DocumentFragment]
Maybe we can do something like
_innerHTML: "$prop,$rule|getPropertyValue"
and let getPropertyValue return raw HTML?

To tired to fix with backing out and such right now.

>> [Object DocumentFragment]
> Maybe we can do something like
> _innerHTML: "$prop,$rule|getPropertyValue"
> and let getPropertyValue return raw HTML?
That was the point! The attached patch (based on Farshid's repository again) includes
that and is working for me. Also css/5461/issue5461.js works again for me.

> To tired to fix with backing out and such right now.
I won't interfere with that. So please go on with that. If my patch is tested well,
we can merge the changes again into master.

Sebastian

Couple of notes from our yesterday's meeting:
- We need to have a separate branch for this issue 
- We need to fix master branch ASAP

Notes:
https://wiki.mozilla.org/Firebug/WeeklyUpdates/2012-10-23

Honza

> issue5880.patch
Right. Remaining:
 - needs HTML escaping
 - I think it's better to make a separate function rather than reusing Str.cropString
 - showInfoTip changes

List of commits that touched relevant pieces of code (mainly just `git log --pretty="format:%ad
%h (%an): %s" --date=short extension/content/firebug/css/cssPanel.js`):
2012-10-22 1da208e (Simon Lindholm): Issue 5204: CSS property disappears when pressing
Esc
2012-10-18 4f0db5a (Sebastian Zartner): Issue 6007 (Add save success indication for
media queries)
2012-10-18 ad4c72e (Sebastian Zartner): Issue 6006 (Editing media queries is broken)
2012-10-03 e006bec (Simon Lindholm): Allow temporary invalid values in selector editing
2012-09-23 ee7e746 (Farshid Beheshti): fixes changing the name of a property with cropped
value
2012-09-21 4404b47 (Farshid Beheshti): fixes the yellow highlighting for the property
name
2012-09-20 771d1c1 (Farshid Beheshti): some changes related to updating css rules names
and a little other changes
2012-09-19 01e3cd3 (Farshid Beheshti): A little changes about some of the mistakes
that need fixing
2012-09-19 215509d (Farshid Beheshti): issue 5862 (crop long css values in style panel)

Some of these may have fixed other issues, and backing them out doesn't feel all that
safe now.

> Some of these may have fixed other issues, and backing them out doesn't feel all that
safe now.
So shall I try myself on that?

Sebastian

Disabled in https://github.com/firebug/firebug/commit/74943f6f2a216bbb38cd22d13f878f54046ed6eb

>> Some of these may have fixed other issues, and backing them out doesn't feel all
that safe now.
> So shall I try myself on that?
Huh? No. Let's just leave the code - I don't know of any bugs, we need it later (kindof
- restoring .textContent makes it a bit redundant but it still separates Controller
and View a bit) and backing it out would undo any fixes made to the CSS code in the
progress.

> Disabled in https://github.com/firebug/firebug/commit/74943f6f2a216bbb38cd22d13f878f54046ed6eb
Ok, fine.

>>> Some of these may have fixed other issues, and backing them out doesn't feel all
that safe now.
>> So shall I try myself on that?
> Huh? No. Let's just leave the code
Well, as I was told by Honza and as I understood from the meeting notes the changes
should be reverted.

> backing it out would undo any fixes made to the CSS code in the progress.
If it's done right (using git revert) the fixes would stay, no?

Anyway, your disabling commit also fixes master. If Honza agrees, I'll create a branch
of it then, so Farshid can merge his changes there and we can finish the work there.

Sebastian

In https://groups.google.com/forum/#!topic/firebug-working-group/EY2WhziH2dA we decided
that this issue is not a blocker.

Sebastian