CSS Test Suite Results: Opera 3.5 for Windows 32-bit

Previous Versions of This Page

• beta 9 1998-??-??
• beta 10 - 1998-11-06
• beta 11 - 1998-11-15
• 3.5 Final - This Version - 1998-11-18

This list of bugs has not yet been updated to include those bugs uncovered by the new version of the CSS Test Suite, released 1998 December 17. The additions to the test suite are significant, and will include many new bug reports.

Bugs

The list that follows is a list of bugs in the browser, that is, what is described is in error. I have also included a few partially supported features as bugs.

• Last run through entire suite: 3.5 beta 9
• Last run through existing bugs to see if they still exist: 3.5 final

1.3

Underlining is ugly. There should be a stable baseline, since text-decoration is not inherited, but simply spans children.

5.2.2

• font-family: cursive doesn't work very well because the default cursive font is "Times New Roman." Things should come out a little better just using the internal TTF font stuff. Of course, the fonts can be fixed up using a very well hidden part of a dialog box (one doesn't expect the CSS fonts to be underneath all the HTML stuff), but most users won't figure that out.
• The font "serif" should be the same as the default font, at least in the initial configuration. (For me, serif was Georgia, and the default was "Times New Roman.")

Note: These bugs are as of beta 11. I did not do a complete uninstall to verify them in the final version.

5.2.5

• bolder should not be any bolder than bold. It is a relative descriptor, but this misunderstanding may have been due to an early incorrect version of the W3C CSS Test Suite.
• Font weight 500 should probably be rendered as bold, as per the example in the CSS2 spec.

5.2.6

ex units are way too big; they are being treated as em (However, they seem to be done right in Section 5.4.1 and other places).

5.2.7

Weight 900 is not rendered as well in this page on italic, larger fonts as it is in section 5.2.5.

5.4.3

The underlining should be a constant color, since text-decoration is not inherited, but spans over children.

5.4.4

The following vertical-align does not work: middle.

5.5.02, etc.

The table width algorithm has problems with CSS margin and padding properties. I also think the TABLEs should, according to CSS, be the whole width of the page, but this is inconsistent with historical behavior.

5.5.22

The top edge of the thin maroon border (the second example) is black (or darker than the rest), as are the top and bottom edges of the double maroon medium border (the sixth example).

5.5.23

• I have display problems when I scroll down using PdDn (not with arrows), when the window size is small enough that the second square is not in the first "page" of the document.
• The width: on TABLE was completely ignored. It seems that width is not supported on tables at all. This is also seen in another page. Some other tests seem to show support for width is very shaky all together. And some others.

5.5.25

I think that according to the CSS2 spec, the vertical placement of the text around the floating images is incorrect.

5.5.25c

• The image in the second example should appear after the paragraph. The same type of problem occurs in the sixth example if one makes the window narrow.
• Due to the above problem, the page is incorrectly given a horizontal scrollbar and made wider than the viewport.

5.6.4, 5.6.6

The list marker images don't have enough padding around them. (There is exactly zero separation between the marker and the text, and that is just a function of the size of the marker.)

5.6.5

With a small font, the text inside the table is carried to a second line even though there is room to widen the table. That is, the bullet is not counted in the line-width calculation. (See my comment on 5.5.02, etc.)

Unsupported Features

• Alternate Stylesheets (1.1)
• A: active selector (2.1)
• background-attachment: (5.3.5)
• all values of display: except for block and none (5.6.1)
• white-space: (5.6.2)

Other Bugs I've Heard About (not from CSS Test Suite)

• Opera does not seem to support background color on a TR. See http://www2.hunterlink.net.au/~ddrge/magazine/mags.html.
• Opera has trouble parsing the W3C Core Styles (it doesn't pick up the A:link and A:visited). This is likely a problem with error handling with CSS2 selectors (and if it is, it would be disobeying the CSS1 rules on error-handling). See my home page.
• clear doesn't always work.
• A border on BODY in a short page should have extra margin-bottom.
• There are numerous vertical align problems.
• There are serious problems in compliance with Section 7.1 of the CSS1 spec.
• Opera does horribly on Ian Hickson's Import Test. I am told this is because it interprets relative URLs in stylesheets in terms of the path of the document intead of the path of the stylesheet.

(Back to CSS Testing Information, David Baron)

LDB, dbaron@dbaron.org