Webrequest

Everything about using Adblock Plus on Google Chrome
Locked
Wladimir Palant

Re: Webrequest

Post by Wladimir Palant »

No, it doesn't collapse - if the web developer defined a particular size for this image then it will still occupy this area. If he didn't then the image will have the size 1x1.
lewisje
Posts: 2743
Joined: Mon Jun 14, 2010 12:07 pm

Re: Webrequest

Post by lewisje »

Wladimir Palant wrote:No, it doesn't collapse - if the web developer defined a particular size for this image then it will still occupy this area. If he didn't then the image will have the size 1x1.
I noticed that you used the (very well-optimized) 1x1-pixel 67-byte transparent PNG

Code: Select all

data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAAEAAAABCAYAAAAfFcSJAAAACklEQVR4nGMAAQAABQABDQottAAAAABJRU5ErkJggg==
rather than the well-known optimized 43-byte transparent GIF

Code: Select all

data:image/gif;base64,R0lGODlhAQABAPAAAP///wAAACH5BAEAAAAALAAAAAABAAEAAAICRAEAOw==
Was this possibly to appease the diehards of software freedom? I mean I personally prefer PNG for most uses, but in a few cases GIF still does better.

It's not a major issue :)
Last edited by lewisje on Fri Oct 14, 2011 2:47 pm, edited 1 time in total.
There's a buzzin' in my brain I really can't explain; I think about it before they make me go to bed.
Wladimir Palant

Re: Webrequest

Post by Wladimir Palant »

I don't think that the "download" size is of any importance here - it's the "optimized" variant mostly for readability reasons. And given that I generally prefer PNG over GIF I didn't see any reason to use GIF here. Especially given that a few years ago I played with the idea of introducing half-transparent "blocked ad" placeholders, something that might still be done at some point.
Guest

Re: Webrequest

Post by Guest »

Wladimir Palant wrote:Especially given that a few years ago I played with the idea of introducing half-transparent "blocked ad" placeholders, something that might still be done at some point.
It would be an improvement over the garish checkerboards or bold black-and-red images that some other filtering solutions use :mrgreen:
Help

Re: Webrequest

Post by Help »

Please fix the Quota_Exceeded_Error 22,, never been fixed even though you say it is.

Adblock Plus for Google Chrome™ (Beta) experimental build1.1.4.703

Google Chrome 16.0.898.0

Screenshot of error. -------> http://temp-share.com/show/HKdPf3ayp

To replicate error, subscribe to at least 3 lists, and hit update 2-5x.

Local Storage is filled.
Chimaira

Re: Webrequest

Post by Chimaira »

Same error than "Help" :)
lewisje
Posts: 2743
Joined: Mon Jun 14, 2010 12:07 pm

Re: Webrequest

Post by lewisje »

Chimaira wrote:Same error than "Help" :)
This occasionally happens to me too :(
There's a buzzin' in my brain I really can't explain; I think about it before they make me go to bed.
adam777

Re: Webrequest

Post by adam777 »

Confirmed here as well.
Causes loss of lists. Meaning, unable to update a list, so it's inactive and needs to be removed and readded.
lewisje
Posts: 2743
Joined: Mon Jun 14, 2010 12:07 pm

Re: Webrequest

Post by lewisje »

adam777 wrote:Confirmed here as well.
Causes loss of lists. Meaning, unable to update a list, so it's inactive and needs to be removed and readded.
OK, I think I know the problem: When lists are updated and placed into Local Storage, the old versions of those lists are not removed; the Chrome Web Inspector shows that the keys in the Local Storage database are the URLs of the lists, so one possible solution could be to remove from Local Storage the keys with those URLs...but maybe it works the way it currently does because once a tab is loaded, the latest copies of the lists at that time might need to remain in Local Storage until that tab changes (like being refreshed or closed or sent to a new address).

I think I'll look through the ABP source code to see why it doesn't just keep the very latest copy of each list in Local Storage.
There's a buzzin' in my brain I really can't explain; I think about it before they make me go to bed.
adam777

Re: Webrequest

Post by adam777 »

Sound interesting, lewisje.
Do tell what you find...
Wladimir Palant

Re: Webrequest

Post by Wladimir Palant »

Ok, I found out that Chrome's "unlimitedQuota" flag doesn't really work (http://crbug.com/58985). Great... Web SQL DB as a standard is dead and will most likely be dropped in a future Chrome version. AppCache cannot be used to store data. So much for storing data in an extension.

Edit: It looks like FileSystem API isn't restricted as far as size goes - I've successfully requested 1 TB of space (not actually available on my disk). And actually managed to write out 80 MB of UTF-8 data. So we should be able to save our data there.

Edit2: And the "fun" part of course is - all FileSystem API operations are asynchronous whereas Adblock Plus/Firefox expects synchronous file read operations...
adam777

Re: Webrequest

Post by adam777 »

Nice to see there's a known cause, at least.
Good luck finding a solution, Wladimir, it really is annoying.
User avatar
pirlouy
Posts: 332
Joined: Sat Jun 10, 2006 2:33 pm
Location: France

Re: Webrequest

Post by pirlouy »

I don't know if it's due to recent Chrome builds, but I have a problem with Youtube and this webm video:
Image

Youtube request are blocked by this filter: ||$third-party,script,subdocument,object-subrequest
It should not be blocked, but even if this filter catched it, there's another filter: @@||youtube.com^ which should solve the problem.

Versions:
Chrome Dev 16.0.912.4
AB+ 1.1.4.703

Edit: forget about it; it's a Youtube "bug"; this Webm video does not work on any browser.
Wladimir Palant

Re: Webrequest

Post by Wladimir Palant »

@pirlouy: Yes, for blocked requests it says "(failed)" rather than "(canceled)".
lewisje
Posts: 2743
Joined: Mon Jun 14, 2010 12:07 pm

Re: Webrequest

Post by lewisje »

It turns out that the important part is in filterupdate.js; I tried adding code to remove all items with the same key as the filter URL being downloaded, but it didn't work. I also read the spec and it looks like it might be a bug in Chromium (LocalStorage is not supposed to allow two key/value pairs to have the same key).

Hmm...I wonder whether it's any harder to use IndexedDB (although it would break compatibility with Chrome versions older than 9).
There's a buzzin' in my brain I really can't explain; I think about it before they make me go to bed.
Locked