<html><body><div style="font-family: verdana; font-size: 12.0px; color: #000000;"><span style="background-color: #ffffff;">Hey all,</span></div>
<div style="font-family: verdana; font-size: 12.0px; color: #000000;"><span style="background-color: #ffffff;"> </span></div>
<div style="font-family: verdana; font-size: 12.0px; color: #000000;"><span style="background-color: #ffffff;">I can at least agree from my side that I never saw any project using BSS and I also did not use it myself.</span></div>
<div style="font-family: verdana; font-size: 12.0px; color: #000000;"><span style="background-color: #ffffff;">The idea is interesting, but I never had time to see if I can add the css2bss step into the pipeline (doing it manually for every change was no option for me).</span></div>
<div style="font-family: verdana; font-size: 12.0px; color: #000000;"><span style="background-color: #ffffff;"> </span></div>
<div style="font-family: verdana; font-size: 12.0px; color: #000000;"><span style="background-color: #ffffff;">I have the same thoughts as John: We may can check if there are low hanging fruits or ideas to improve the default CSS parsing. Maybe we then don't need the BSS feature.<br>It is somewhat similar to the discussion whether to use FXML or writing the UI in plain Java code. <br>Same here: If there is no big slowdown/disadvantage, I don't see a problem.</span></div>
<div style="font-family: 'verdana'; font-size: 12px; color: #000000;"><span style="background-color: #ffffff;"> </span></div>
<div style="font-family: 'verdana'; font-size: 12px; color: #000000;"><span style="background-color: #ffffff;">-- Marius</span></div>
<div style="font-family: verdana; font-size: 12.0px; color: #000000;"><span style="background-color: #ffffff;"> </span></div>
<div id="sub-body-container" style="margin: 10.0px 5.0px 5.0px 10.0px; padding: 10.0px 0.0px 10.0px 10.0px; border-left: 2.0px solid rgb(195,217,229);">
<div style="margin: 0.0px 0.0px 10.0px;">
<div><strong>Gesendet: </strong>Samstag, 20. Dezember 2025 um 05:56</div>
<div><strong>Von: </strong>"Michael Strauß" <michaelstrau2@gmail.com></div>
<div><strong>An: </strong>openjfx-dev <openjfx-dev@openjdk.org></div>
<div><strong>Betreff: </strong>Removal of binary CSS (BSS)</div>
</div>
Hi all,<br><br>I ran some numbers on the JavaFX binary CSS (BSS) feature. For the<br>measurements, I used CssParser to parse modena.css and modena.bss into<br>a Stylesheet instance.<br><br>On a cold start, parsing takes ~150 ms for the CSS file and ~100 ms<br>for the BSS file (about a 50% improvement for BSS). Using JMH to<br>benchmark warmed-up performance, I get ~4.4 ms for CSS and ~1.7 ms for<br>BSS (about 2.6x faster for BSS).<br><br>Those results are a meaningful improvement, but the downside is the<br>added complexity: the binary path introduces non-trivial code that has<br>to be maintained over time, and it tends to slow down or complicate<br>new feature work significantly.<br><br>My current hypothesis is that this is largely irrelevant for most<br>applications in practice, because modena.css is likely the only<br>stylesheet that ever benefits from the binary path. BSS doesn't come<br>"for free"; projects have to run css2bin as part of their build, and I<br>suspect most applications won't do that. So in the common case, the<br>only stylesheet parsed via BSS is Modena, where both modena.css and<br>its prebuilt modena.bss ship with the JavaFX runtime.<br><br>Maybe in 2010, BSS was a meaningful improvement for mobile devices.<br>But today, these gains are dwarfed by overall JavaFX and JVM startup<br>time. Removing the binary path would free up development resources for<br>features that actually matter for JavaFX applications.<br><br>Thoughts?</div></body></html>