Opera Core Concerns

Welcome! It's great to see you guys finally on the blogosphere. I wish you much success.

Please fix blog header, because currently there is no way you can go to home page of this blog. Simply put a link to header of this blog.
Desktop Team hads this problem too, but they fixed it already.

Publish a ne blog is a good idea.Congratualtions.

I know previously pretty printing of javascript (Function.toString) was removed for performance reasons. Any plans for it to return in a similar way of the new performance tweak for stack traces?

It`s allways good to get as much informations about my favorite browser as I can get. Gonna check this blog out regulary. Thanks

Cool blog. A thing I was wondering though; the logo used, the cube thingy. Would it be possible to release it as wallpaper? I would love to see it as a wallpaper with 1650x1080 resolution. It looks awesome!

Congratulations with the new blog!

We heartly welcome u home at our blogosphere. We hope u will provide us with user-friendly advances to the right direction.

I'm also interested in when/if opera will be upgrading from javascript 1.5. Both 1.6 and 1.7 have very useful features in them. 1.6 would improve performance on many websites/web apps if they were defined natively. For instance, Array.(filter|some|each) are defined by most js libraries but when the functionality is already built in, they generally don't override them. All native functions are faster than their client-side counterparts in my experience. This would make sites run/load faster in Opera if it caught up with some of the newer things on the block. Js 1.7 offers lots of conveniences like grabbing numerous values from one statement

var parsedURL = /^(\w+)\:\/\/([^\/]+)\/(.*)$/.exec(url);
if (!parsedURL)
  return null;
var [, protocol, fullhost, fullpath] = parsedURL;

For reference:

Browser - version supported
---------------------------
Firefox 2 - 1.7
Firefox 3 - 1.8

Safari 2 - 1.5
Safari 3 - 1.7

Internet Explorer 6 - 1.3
Internet Explorer 7 - 1.3

Icab 3 - 1.6

Opera - 1.5

I do realize that that not all of the extensions added by mozilla are standards yet.

Let me know if there is a better place to direct my questions. I have lots of core concerns and questions that I have had no one to talk to about. It would be cool if there was an irc channel to hang out in to talk about core stuff... like there is for the desktop team... or if any core devs hung out regularly in #snapshot. To summarize, anything making this more social is good.

I also have a laundry list of core bugs related to the js engine and its error reporting. These also negatively impact the dragonfly project since it relies on the same error text.

There is a regression of a previously fixed regression in the latest build (10024). I can file it again or perhaps you wish to reopen the previously fixed bug #283596.

Sorry for being a little eager. It is hard to get an audience of core devs in my experience... even an audience of one.

E-mail address removed by moderator.

Is/would it be useful for me or people to point your attention to bugs? For instance, I have a bug about return being allowed outside of functions. I'm not saying it should be fixed already but i realize that some things can get lost in all the hustle and bustle.I find that most of I find that most of my bug reports are against core functionality.

Although userjs-related, I'm pretty sure this would have to start at the core level. Will userjs ever receive elevated XMLHttpRequest privileges so it can do cross-site communications?

Is this blog abandoned? Almost 3 months (in 9 days) with no posts. Just curious. Also, I do know that a lot of people are on holiday... but I didn't know Opera "shutdown" essential.

Am i asking the wrong questions? Am i asking in the wrong places? I've made 4 posts with multiple inqueries each and 3 of my posts are nearly 3 months old with no response. Although it would suck, I would accept "we can't talk about that, nda, etc." That is better than simply being ignored.

EDIT: Will oncontextmenu be supported as web applications are much more prevalent now?

@fearphage: what makes you think that this is a Q&A? Did they ask you to submit questions? Use the forums or something if you want to talk about bugs. Don't post off topic stuff here.

Originally posted by GoJoeGo:

Don't post off topic stuff here

You mean off topic as in things that only pertain to core functionality in the core blog? I'm posting here because this is the only place where there is an audience of core devs/staff. If they were on irc, I'd ask there. Did anyone say they would not answer questions? If you are under some assumption that my post is off-base, then yours is just as far off-base, sir. Unless you have talked to the core staff, perhaps you should keep your assumptions and condescending remarks to yourself.

Another major issue I have with the error console:

JavaScript - http://www.nytimes.com/ads/common/embed3.js
Linked script not loaded
JavaScript - http://ad.yieldmanager.com/st?ad_type=ad&ad_size=336x280,300x250&section=394773
Linked script not loaded
JavaScript - http://partner.googleadservices.com/gampad/google_service.js
Linked script not loaded
JavaScript - http://edge.quantserve.com/quant.js
Linked script not loaded
JavaScript - http://ad.doubleclick.net/adj/fmpub.gigaom/;sz=728x90;tile=1;fmzid=382;ord=3059353567659855?
Linked script not loaded

Do you know why those scripts weren't loaded? because they are blocked by content block. My console is full of this information basically telling me that content block is working. To make matters worse, they are at the highest level possible (error level) so I can't even filter them out. It would be quite useful if not-loaded scripts could be checked against content blocked and sent as message_level events instead of error_level so they can be more easily ignored. Currently it makes debugging difficult as you have to filter through all the non-errors along the way.

Are these issues bug-able?

I just wanted to make it known that I did notice the update to security errors in peregrine alpha:

JavaScript - http://localhost/
Inline script thread
Error:
name: ReferenceError
message: Security error: attempted to write protected variable 'domain'
stacktrace:   Line 2 of inline#1 script in http://localhost/
    document.domain="newspaper.com";

Thanks a lot. Keep them coming.

Opera was idling at 50% cpu and acting erratic in general (not loading style sheets and whole pages sometimes). I closed Opera with the close button and it began the close sequence. CPU usage dropped to under 10% for a while and then a few minutes after it should be closed it shoots back up to 40-50% (still in the closing process). My curiosity led me to open procmon to see what the heck opera was doing when it should have closed minutes ago. Opera was going through and touching the cache files.

Backstory
I have had Opera open for roughly a week with anywhere from 20 - 200 tabs open

• Delete cache on exit - disabled
• Show email client - disabled (sometimes it matters)

@staff: Can you please explain what routines opera runs through after it is closed?

Side note: (Possible unrelated) Opera crashed during the file thrashing process. I already reported the crash-on-exit bug a while back as DSK-244636. I see others have filed this bug as well: DSK-246921 & DSK-248169

Can you please give any details about why opera is the only browser that still has "click to activate" for plugins? Is this a bug? Will it be fixed?

@haavard: Thanks for that. I've read that the patent doesn't apply to open-source software and Microsoft paid their way out. Are there any plans to pay or any other way to natively get around this inconvenience?

The current state of the core's error reporting needs a lot of work to be useful and consistent. The properties provided from errors can not be relied on. Too often, errors are thrown and the error.name is literally "Error" or not set at all. I filed a related bug as #323888. That is not remotely helpful. Based on moz and webkit, the convention (i'm not sure if there is a standard for this) is that the name of the error is either the type of the error or the violation name: TypeError (throw new TypeError()), INVALID_STATE_ERR, INVALID_MODIFICATION_ERR, etc.

In addition to that, the message property of errors still generally tend to contain the stacktrace although there is a separate stacktrace property (at times). I filed this bug as #DSK-245714. Although some messages have improved (security errors), there is still room for more.

Can you talk about google gears support (or the lack thereof)? How is it that it works in the mobile but not the desktop? Aren't they based on the same core? Does it need something beyond core support? Any information would be appreciated.