Forum Discussion

MissyQ's avatar
MissyQ
Former Employee
Aug 19, 2020

Announcing a new way to paste URLs, Link format!

Recently, we wanted to dig into how users feel about copying and pasting a URL from their browser. We looked to understand what they do when they copy and paste URLs, as well as the pain points they ...
rshupak's avatar
rshupak
Iron Contributor
Oct 01, 2020

JaredB81 One bug I reported via feedback is when you right click on Articles above and copy link.

 

 

   This produces Articles (microsoft.com).  Where is the " (microsoft.com)" text coming from?  Click on the Microsoft Edge Insider link that is adjacent and it produces Microsoft Edge Insider without the extra text.

 

   The the Link Preview Format clipboard data JSON has a different title still.

 

{"description":"","domain":"microsoft.com","favicon_url":"https://techcommunity.microsoft.com/html/assets/favicon2.ico",
"image_url":"https://cdn.techcommunity.microsoft.com/lithium/TechCommunity-DefaultHero-2.png","preferred_format":"text/h
tml;content=titled-hyperlink","title":"Articles - Microsoft Tech Community","type":"website","url":"https://techcommunit
y.microsoft.com/t5/articles/bd-p/EdgeInsiderAnnouncements"}

JaredB81's avatar
JaredB81
Former Employee
Oct 01, 2020

Thank you, rshupak.  As always, you have a very keen eye.  We append the domain when it isn't included in the title text so that it's clear what the content is related to.

 

Do you have a preference between the text in your example (with the appended domain) vs. the Link text as specified in the Link Preview Format clipboard data's JSON payload?  

 

Thanks,
Jared

  • rshupak's avatar
    rshupak
    Iron Contributor
    Oct 01, 2020

    The title in the JSON is the title I would see if I navigated to that page and copied from the address bar.  I do not see any network traffic in the dev tools and haven't looked to watch network traffic from my PC but I wonder if you are calling an online service which takes a URL and returns that info.  I know Bing has such a service which some Microsoft products and services use.  Do you know?

     

    I do not like the domain being added.  This breaks consistency that when I copy a link from one document to another that it isn't copied as I see it.  Compare this to the same scenario when copying a link in Word or an Outlook email.  Right click on a link, pick Copy hyperlink, then paste.  It matches.  

     

    This also creates silly cases such as when you visit https://www.bankofamerica.com/ and copy from the address bar.  The title in the HTML format is "https://www.bankofamerica.com/" because you don't recognize "Bank of America" as "bankofamerica".  You could fix this but it isn't easy to consider all cases of how names map.  Bank Of America also uses "bofa.com".  Citibank uses citibank.com and citi.com and others.  Next, consider IRIs like https://www.смарт-скиллс.рф.  The title copied here is "https://www.xn----7sbzhgab7ageef.xn--p1ai/".  Do you think that the punycode domain name is useful?  Do you look for either the IDN or the punycode in the title?

     

    Editing to add one more case I forgot to mention.  In some cases the domain name doesn't quite match the localized version of the name.  Consider https://lt.wikipedia.org/wiki/Microsoft which is titled as "https://lt.wikipedia.org/wiki/Microsoft" because the localized Vikipedija doesn't match Wikipedia.

    • rshupak's avatar
      rshupak
      Iron Contributor
      Oct 02, 2020

      JaredB81 I should point out that these issues manifest with localized Microsoft.com.  As an example, https://account.microsoft.com/privacy/?lang=ru-RU has the title "https://account.microsoft.com/privacy/?lang=ru-RU" because they transliterated Microsoft into Cyrillic.  I expect that this will be a very common issue with non-Latin language pages hosted on sites that use non-IDN domain names.  

      • rshupak's avatar
        rshupak
        Iron Contributor
        Oct 07, 2020

        JaredB81 Here is another bug.  It takes some setup to get to a case I know fails.  Open https://www.bing.com, type into the search box "История Киева." with a trailing period.  Then select all but the period.  This is necessary to work around another Edge bug.  Right click on the selection and pick Search the sidebar for "История Киева".  In the sidebar, click on IMAGES to display this pivot.  Right click on the following image and pick Copy image link.

         

         

        The clipboard will contain the following formats.  The Cxxx values will likely vary.  The bug is that the last three are all empty zero terminated strings, i.e. a single byte of zero.  I have not tested to see if this is a problem because the image is a data URL or because the data URL is very long (12,056 characters) and some part of the code to generate the other formats has a URL length limit (e.g. INTERNET_MAX_URL_LENGTH) and silently fails while failing to suppress adding the failed result on the clipboard.

         

        000D Unicode Text
        0010 Locale
        0001 Text
        0007 OEM Text
        C0DA HTML Format
        C2C1 Link Preview Format
        C2C2 Titled Hyperlink Format

         

        I sis submit this via the feedback mechanism.  In my feedback I questioned whether Copy image link should be exposed at all for data URL encoded images as Copy image is more appropriate I would think.

         

Resources