Developers › Desktop Team blog
Core update with better CSS unit precision
New Opera Next snapshot that brings Presto all the way up to version 2.10.269.
We’ve introduced a single unified approach for dealing with layout’s fixed point values (lengths, font sizes). The new approach squashes our old limiting and imprecise rounding to the nearest integer for percentage and em values.
There are 69 other entries in the changelog, so go crazy!
Known issues
WARNING: This is a development snapshot: It contains the latest changes, but may also have severe known issues, including crashes, and data loss situations. In fact, it may not work at all.
Download
We’ve introduced a single unified approach for dealing with layout’s fixed point values (lengths, font sizes). The new approach squashes our old limiting and imprecise rounding to the nearest integer for percentage and em values.
There are 69 other entries in the changelog, so go crazy!
Known issues
- DSK-358417 (Extension preferences and localStorage lost on upgrade)
—work around by uninstalling extension, restarting Opera, and reinstalling extension - Very fast page loading
- Adding bookmark from star menu freeze or crash
WARNING: This is a development snapshot: It contains the latest changes, but may also have severe known issues, including crashes, and data loss situations. In fact, it may not work at all.
Download
Changelog
- Stability improvements
- Update the Media Queries implementation to latest draft
- Increase and unify precision for most CSS length properties
- Minor hardware rendering performance improvements
- CORE-42332 (Frames not printed)
- CORE-40370 (Unstyled content while page with multiple stylesheets)
- CORE-43099 (New XMLHttpRequest responseType "json")
- CORE-43117 (External JSON.stringify() testsuites failures)
- CORE-28562 (Does not handle dates past 2038 even when time is 64 bit)
- CORE-43154 (XMLDocument has no elementFromPoint)
- CORE-43146 (In regex \s incorrectly does not match BOM)
- CORE-43119 (validation error on submitting form with input type="number" with value 5210000010001001)
- CORE-43317 (Upgrade SQLite to 3.7.9)
- CORE-43265 (Gstreamer initialized on every page load)
- CORE-1588 (initEvent on dispatched event must have no effect)
- CORE-43247 (race condition leads to same-origin-policy bypass)
- CORE-43232 (Implement getUserMedia options object)
- CORE-42697 (Bad webfonts cause never-stop-loading (dn.se never finishes loading))
- CORE-42761 (Hidden SVG element prevents hover of underlying element in Raphael.js)
- CORE-40831 (Web SQL database - big int column not being returned correctly)
- CORE-43209 (Wrong status code is given in the onreadystatechange handler for XMLHttpRequests (XHR2, CORS) to targets from other origin)
- CORE-43260 (bgcolor=transparent (on ) is not ignored)
- CORE-42545 (crash when running IDNA2008 tests)
- CORE-43519 (Opera doesn't prompt for user/pwd for proxy https)
- CORE-35564 (reload (or refresh from cache) of the document with objects does not preserve position)
- CORE-43550 (The new ecmascript timer code can hang)
- CORE-43372 (XHR POST with explicit content-type and no entity body fails)
- CORE-43492 (MediaStreamOptions unsupported (DOM.Media.getUserMedia))
- CORE-43493 (Crash on reloading of multiple origin camera test)
- CORE-33779 (Character encoding of cached stylesheet not re-evaluated for new referring document)
- CORE-42348 (Replacement character shown in Mercado Libre search - SCRIPT tag's charset attribute ignored when set from script)
- CORE-43365 (synthetic keypress events exhibit flaky failures)
- CORE-43423 (Text inside button with display:block is selectable)
- CORE-42748 (Opacity ignored for inline-block transform rotated to a certain angle)
- CORE-38167 (if max-width is used for absolutely positioned element with zero offsets and auto margins, the margins are incorrectly set to zero)
- CORE-43398 (Missing 64-bit architecture information in the User Agent)
- CORE-42133 (SVG mask transparency inserted via javascript does not work)
- CORE-43478 (constants on DOMException interface have writable and configurable set to true)
- CORE-43480 (constants on Node interface are writable and configurable)
- CORE-43357 (XHR firing error event on response status codes != 200)
- CORE-43467 (drawImage(video) aspect ratio correction must use float arithmetic)
- CORE-43297 (CSS cursor rule is not working on <input> elements)
- CORE-15578 (Number.prototype.toString() not accurate for large numbers and radix not equal to 10)
- CORE-43444 (selectionStart/selectionEnd after value change don't work correctly in a text field)
- CORE-42043 (Improve parseInt performance)
- CORE-43589 (Severe slowdown with urlfilter.ini and data:image)
- CORE-7897 (Line numbers of stack traces of inline elements are reported relative to the SCRIPT tag)
- CORE-21925 (JavaScript -> <textarea>s -> onscroll event inoperative)
- CORE-42347 (PageDown / Up scroll amount should depend on window size)
- CORE-43672 (Crashes on pages with invalid character encodings)
- CORE-43466 (Animated SVG with display: none trigger slow repaint of the whole view)
- CORE-43564 (Setting "border-radius" to "inherit" fails)
- CORE-43092 (Make Array.prototype.{join, concat} faster)
- CORE-43574 (Calling preventDefault() on non-cancelable events should have no effect)
- CORE-8358 (innerHTML in XML doesn't output namespace declarations correctly)
- CORE-41036 (Fix for crashes on 64-bit)
- CORE-42504 (Linear gradient is visually broken)
- CORE-43597 (Incorrect cache resolution over string values)
- CORE-43324 (CSS import test takes too long to load)
- CORE-43716 (Dropdown menu on blocket.se impossible to use - pops up when it should close.)
- CORE-42674 (Expose readonly properties per WebIDL (like Event.target))
- CORE-43302 (load events fire on LINK elements pointing to existing local resources (link rel="stylesheet" href="file://..."))
- CORE-43035 (Origami 3D Engine is slow in Opera because of slow blending)
- CORE-42466 (Scrolling list items too fast on Mac)
- CORE-43683 (Canvas: shadow prevents subsequent fill)
- CORE-38715 (Incorrect 'inset' box-shadow if border-top-width=0 xor border-left-width=0)
- CORE-23426 (Punctuation class Pc should not be included in ::first-letter)
Readers' comments
Comments left by My Opera members under the original post: the first 50, as archived.
About this article
| Published in | Desktop Team blog |
|---|---|
| Date | February 28, 2012 |
| Author | Daniel Aleksandersen |
| Tags | wahoo, performance, opera-next, sixty-nine, core |
| Original | Internet Archive |
KilokilozaTuesday, February 28, 2012 4:56:07 PM
Thank you, looking good!
Very impressive changelog.
BakLANTuesday, February 28, 2012 4:59:28 PM
ThangarajthangarajjTuesday, February 28, 2012 5:05:38 PM
Don't laugh, this is my highest yet!
Rob SegalrobmsegalTuesday, February 28, 2012 5:08:26 PM
I'm still getting a crash when extension buttons on the toolbar and freeze when clicking the bookmark star.
AvolaTuesday, February 28, 2012 5:18:56 PM
Edit: Holy crap! Pages don't shift to the left when they finish loading from cache anymore! Thank you, that's annoyed me since 10.50 alpha I think.
Ayrton Nádgelayrtonnadgel92Tuesday, February 28, 2012 5:25:13 PM
Егорtermi1uc1Tuesday, February 28, 2012 5:25:32 PM
Crash every time I open SpeedDial. Random crash when open any site
Win7x64
Open ~30 tabs with 3 tab stack
10 crash reports sending
Downgrade to previous snapshot
Update:
Crash when update from previous snapshot.
No crash with clear install without Opera Link (I fear turn it on) and custom search engines
Update 2:
Visual artifacts with scrolling any page on top and bottom
SamHuRRaCaNeTuesday, February 28, 2012 5:27:57 PM
Originally posted by Avola:
Yep, confirmed this.
blackcoderTuesday, February 28, 2012 5:28:18 PM
Christophchristoph142Tuesday, February 28, 2012 5:30:27 PM
Originally posted by HuRRaCaNe:
confirmed Win8 DevP x64artemeasTuesday, February 28, 2012 5:32:43 PM
MyOpera team, please fix this!fearphageTuesday, February 28, 2012 5:34:52 PM
I'm loving the all core fixes. SEXY!
artemeasTuesday, February 28, 2012 5:35:21 PM
Originally posted by Avola:
confirmed ArchLinux x86_64Damian SkoczylasdskoczylasTuesday, February 28, 2012 5:37:26 PM
Vista 64bit SP2, Geforce 8800GT (295.73)
sebtTuesday, February 28, 2012 5:37:45 PM
The annoying "flashing" during page transitions seems to be gone for the majority of the sites given as examples of the problem
However, on http://demo.yootheme.com page transitions are still ugly and the page is displayed momentarily before CSS is applied. On initial load of a new page, DIV layout is and menus show incorrectly. Reloading the same page with a click fixes the layout.
So a partial win. Page transitions are generally much nicer now as the screen is not blanking each time before updating. It still seems however that the config option "redraw after xx seconds" is not being applied all the time.
Perhaps you could look at demo.yootheme.com to address these issues; they're a good example of what I'm seeing on a number of sites using complex layouts, jquery, gzipping and minifying before loading.
Seb
Shane BundyoperabakerTuesday, February 28, 2012 5:38:20 PM
Does this mean what I think it means? I know some people on the forum had this issue and now the fix is here. Thank you!
On a side note, why is DSK-357882 absent from this build?
BS-HarouTuesday, February 28, 2012 5:43:38 PM
Increase and unify precision for most CSS length properties
CORE-43099 (New XMLHttpRequest responseType "json")
Love this build! And I do really confirm very fast page loading.
(but what is the point of CORE-43232 & CORE-43492 without webcam support?
Shane BundyoperabakerTuesday, February 28, 2012 5:55:12 PM
minhoTuesday, February 28, 2012 5:56:32 PM
Put the mouse pointer on "Testes interativos", "Rituais", "Eventos" or "Atendimentos" (on top).
The page is moved!
win 7 x64
PhileeTuesday, February 28, 2012 5:58:55 PM
I should have read the known issues first.
I've lost all my saves sites in the Tab Vault extension.
Anyone knows if they are still somewhere located?
Cache, backup file,...?
Rus-landTuesday, February 28, 2012 6:00:10 PM
eliotcougarTuesday, February 28, 2012 6:02:19 PM
FINALLY!!! Opera passes Acid1 test...
ClashCityRockerclashcityrockerTuesday, February 28, 2012 6:02:34 PM
"Implemented HTML5 drag and drop. (Presto/2.10.270)"
:-)
beBossTuesday, February 28, 2012 6:05:37 PM
Runtime Error!
Program: C:\Program Files\Opera\opera.exe
R6025
- pure virtual funciton call
XombiemubaidrTuesday, February 28, 2012 6:05:58 PM
Originally posted by clashcityrocker:
Ohhh yesss!
known issues: Very fast page loading.
AvolaTuesday, February 28, 2012 6:08:01 PM
Edit: lol, it crashed when I was submitting this post too.
Manoj ChandrasekharanmachsTuesday, February 28, 2012 6:09:21 PM
Originally posted by termi1uc1:
Crash on SpeedDial confirmed. Fresh, standalone installation on Windows 7 64bit
It runs fine on the first run when opera.com opens directly, but when I click on new tab, SpeedDial appears and Opera crashes.
Also it crashes on startup automatically, if the opening page is SpeedDial
BTW, impressive changelog...too bad will have to skip this build
XombiemubaidrTuesday, February 28, 2012 6:11:52 PM
Originally posted by operabaker:
D1sasterp1eceTuesday, February 28, 2012 6:13:45 PM
Aside from that, I still occasionally get the aforementioned runtime errors and crashes. I can also confirm these just mentioned visual artifacts - the faster I scroll, the bigger screen area gets 'smudged'.
I guess I'll be sticking with 11.62 for a little more time.
Muneer Hassan Abdullamha2999Tuesday, February 28, 2012 6:15:40 PM
eliotcougarTuesday, February 28, 2012 6:18:26 PM
minhoTuesday, February 28, 2012 6:19:33 PM
Originally posted by termi1uc1:
Unconfirmed.
win 7 x64
Shane BundyoperabakerTuesday, February 28, 2012 6:21:38 PM
Originally posted by mubaidr:
That is exactly what I'm waiting for. This build makes Opera faster in many synthetic benchmarks but for general browsing the DSK-357882 fix makes all the difference. I'm stuck on the software-backend because my Intel Celeron laptop is ancient with its GME965.FhuryTuesday, February 28, 2012 6:25:18 PM
Very fast page loading
Hihi, top!
KilokilozaTuesday, February 28, 2012 6:32:35 PM
Opera.exe
R6025
- pure virtual funciton call
Unstable... alas... crashes with above error all the time.
Back to using Opera labs 64bit snapshot again.
/sad face
EDIT: Win 7 64 bit - clean USB install.
MeKsevioTuesday, February 28, 2012 6:35:57 PM
Satan D. LucifersupertrolTuesday, February 28, 2012 6:37:52 PM
Originally posted by mubaidr:
Is that the fix from 11.62?(i think so)
but this build i notice a benchmark improvements in my system
PK benchmark increase in 200+ points(previous wahoo build score 1700+ but with this its 1900+) (only software not HWA enabled)
sunspider tests improve little bit by 40 points.
but i like the fix from 11.62 snapshot though(i need more network core fixes not benchmark...i already know who is the speed king in my system and i am using it currently
one thing i notice
why is this opera:config#Performance|HTTPErrorStrategy not present in this build?
11.61 has it.
minhoTuesday, February 28, 2012 6:39:30 PM
win 7 x64
Harlekin MondharlekinmondTuesday, February 28, 2012 6:39:38 PM
newscpqTuesday, February 28, 2012 6:42:29 PM
Originally posted by artemeas:
incredible, really, I was going to switch to other browsers, but now Opera is incredibly FAST!!!Originally posted by Daniel:
CONFIRMED!!!.
Shane BundyoperabakerTuesday, February 28, 2012 6:44:31 PM
Originally posted by supertrol:
Yes it is, and I'm surprised it wasn't dropped in Wahoo first.
This build improves performance on synthetic benchmarks like SunSpider and Peacekeeper but it's DSK-357882 that boosts general browsing speed for me.
As for the Visual C++ Runtime errors I've only seen that error once when I changed Opera's skin and closed it afterwards.
Ned the NaniteNedTheNaniteTuesday, February 28, 2012 6:52:05 PM
Originally posted by kiloza:
Same here in Win XP — I get about a minute of browsing time usually before it pops up.
Strangely enough, if you just ignore it Opera continues to function fine.
Pablo BellucciampkstroffTuesday, February 28, 2012 6:54:08 PM
Patkos CsabapatkoscsabaTuesday, February 28, 2012 6:57:41 PM
GL_VENDOR NVIDIA Corporation
GL_RENDERER GeForce 8600 GT/PCI/SSE2
GL_VERSION 3.3.0 NVIDIA 290.10
However, HWA is still unusable. It's somewhat faster but still it's just too unresponsive. For example I double-click the tab-bar ... new tab appears after about 1 second. I start typing something in addressbar, popup appears after about 2 seconds. Scroll is interrupted sometimes.
Still, these times are much faster and better than in the previous snapshot where I waited for a new tab even 3 seconds sometimes.
So, keep up the good work on HWA. I can't wait it to be as fast as software backend. Also stability is still an issue (sent automatic crash report
PengePenge4Tuesday, February 28, 2012 7:08:16 PM
Originally posted by Philee:
%appdata%\Local\Opera\Opera\widgets\
The easiest way:
1. Open the wuid-********** folders.
2. Open the prefs.dat files in any simple text editor (Notepad, Total Commander Lister, etc.) to identify the name of extension.
3. Backup
senseysensorTuesday, February 28, 2012 7:08:31 PM
King CrimsonTuesday, February 28, 2012 7:09:51 PM
1. still crash on exit (the same old runtime error!)
2. can't read name products (on mid-top) here: http://www.panasonic-batteries.com/eu/products/primary-batteries
Win7 x64
СветославskodjamanovTuesday, February 28, 2012 7:10:33 PM
Originally posted by beBoss:
+1Алексейre-readTuesday, February 28, 2012 7:11:34 PM
Guiot SidneyFirefly74940Tuesday, February 28, 2012 7:13:10 PM