User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.4b) Gecko/20030509 Build Identifier: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.4b) Gecko/20030509 When using relative units to set a border-width, like 'border-width: 0.1em' it seems that the conversion from em to px is done more than once, resulting in different border-widths sometimes. Have a look at the sample page. Reproducible: Always Steps to Reproduce: 1. 2. 3. Expected Results: The converted pixel-value corresponding to a relative value should be locked so that it's always the same no matter how many times or where it is used.

->layout B&I

> it seems that the conversion from em to px is done more than once It's done once, but the result is a fractional number of pixels. We lay out using those fractional numbers, but then your monitor is not capable of displaying sub-pixel-width lines, and rounding occurs.... When rendering to a device capable of sub-pixel display (eg a printer), this is a non-issue.

(In reply to comment #3) > When rendering to a device capable of sub-pixel display (eg a printer), this is > a non-issue. I would generally agree. But would it be possible than to make such a screenshot? I took on on a CRT and one on a TFT, although that doesn't matter. It might matters that these 2 were taken on different graphic cards. Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.9a1) Gecko/20060531 Minefield/3.0a1

This WFM on Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.9b2) Gecko/2007121016 Firefox/3.0b2

The borders are equally sized starting with Firefox 3.0 The reference uses pixel values, is that reliable in this case?

Comment on attachment 357402 [details] [diff] [review] reftest You should: * specify the font-size on body, since the default can vary * drop the test in 'ex', since that can vary depending on what the default font is With those changes, r+sr=dbaron.

(In reply to comment #8) > (From update of attachment 357402 [details] [diff] [review]) > You should: > * specify the font-size on body, since the default can vary Done. I used 16px as that's the browser default. > * drop the test in 'ex', since that can vary depending on what the default > font is Done. The "new" test still fails in Fx2 and passes in Fx3+. > With those changes, r+sr=dbaron.

Pushed the test: http://hg.mozilla.org/mozilla-central/rev/723ea3c0719f