Hyperlink removes selected text

[DrChiper]

When selecting text and using the new hyperlink function, the selected text is actually removed and replaced by just the hyperlink. I do not think that that is the intended behavior. Perhaps I did something wrong?

[Divingduck]

This happen to me as well.
As work around, I do only point the place for the Link and add the text after inserting the hyperlink. However, it will be more comfortable to add the link to a text in the file, if there is a selection.

[kovidgoyal]

That is the intended behavior. It allows you to easily replace a hyperlink. If you want to insert it, just dont select any text.

[DrChiper]

Thanks for clearing that up

[davidfor]

That isn't the behaviour I expected. Most places that I have seen something like this, including this forum, use the selected text as the link text. That makes it very easy to write the text and then add the links afterwards.

And I'm not sure how it makes it easy to replace the hyperlink. If I select a hyperlink in the text and press the button, it replaces it including the link text. What am I missing?

[BetterRed]

Originally Posted by davidfor

... Most places that I have seen something like this, including this forum, use the selected text as the link text.

FWIW - its also how it works the calibre Comments Editor

Select the desired text, click the link widget in the toolbar, paste url into the Enter URL text box, press enter. The desired text will go blue underlined.

If you click on blue underlined text and its an http link then relevant page opens in your browser, if its an FTP link..., if its a file link...,

BR

[theducks]

I would have expected the/any selected text to have become the label (visible portion) of the hyperlink . After all you need something there to select

I can see an empty-(selection) string assuming the Link value (the action that happens when you use the 'Insert Link' tool on MR) for the label in this case.

[DrChiper]

Well, you need also something to create a link target.
Many times I create double linked hyperlinks for reference purposes as not all E-readers have automagical return support (like for instance the PRS-Tx):

• click the ref-link and you jump to a reference,
  

  <a id="footnote1"></a><sup><a href="footnotes.xhtml#bookmark1">1</a></sup>

• click the reference and you jump back where you started.
  

  <a id="bookmark1"></a><sup><a href="Chapter0X.xhtml#footnote1">1</a></sup>

They are quite handy, but must always be manually crafted

[kovidgoyal]

This is an html source code editor, there is no guarantee that the selected text does not contain other tags. You cannot just assume it is text a plunk it in the middle of an <a> tag.

There are three possible behaviors for the insert link action when text is selected:

1) Delete the selected text and insert the <a> tag (this is what currently happens)

2) Ignore the selected text and insert the <a> tag either before or after it depending on the current cursor position at the start or end of the selection.

3) Wrap the selected text in <a> tags.

1) and 2) have the advantage that the insertion of the <a> tag can be guaranteed to not produce invalid HTML. Trying to make that guarantee with (3) is very difficult. The best that you could probably do is detect a potential problem and bail with an error message.

[DrChiper]

Which is quite similar as what you mentioned in a previous discussion concerning the <h1> tags. I already assumed that this would be the reason for the implementation.

[mrmikel]

Since it is an html editor, one would hope they would know what they are doing. They can create much worse messes than the one cited using any html editor. My request is wrap the selected text and hope the person using the program is somewhat knowledgeable. The length of the wrapped text is likely to be a word or two, generally, or a title of something.

As it stands now, I will click at the beginning of a phrase, click the hyperlink button and then cut the ending tag off and move it to the end of the phrase.

[theducks]

We are using a code editor as you pointed out.
I would agree 100% with your plan if the user was working on the WYS(WYG) side

I guess the question is where do you stop protecting the user from producing invalid code?

I can already delete any part of tag code, leaving it invalid.

I don't know how hard it would be to Sanity validate the selection and report:

"Selected text contains unbalanced tag pairs"

[DrChiper]

Originally Posted by mrmikel

.. one would hope they would know what they are doing.

where do you stop protecting the user?

Although using the edit book function is not for the very faint of heart, it is not rocket science either
I think we can safely assume that users made a educated choice when starting an edit book session. But you can always throw in a warning on opening an e-book, like: You must be really, really know what you are doing before continuing.

(I think this would be overkill, but hé, who am I to judge? )

[kovidgoyal]

I have absolutely no problems with not protecting the user from doing things that directly create invalid code. I do have a problem with the user asking the editor to perform some automated action that leads to invalid code. That is something I try to avoid as much as is reasonably possible.

Things the editor does as opposed to the user doing should try to not result in invalid code, as much as possible.

[DrChiper]

Originally Posted by kovidgoyal

I do have a problem with the user asking the editor to perform some automated action that leads to invalid code. That is something I try to avoid as much as is reasonably possible.

That is a solid and sound design decision I fully agree with