E17 benchmarks

Brian J. Tarricone bjt23 at cornell.edu
Thu Jun 9 20:46:56 CEST 2005

Hash: SHA1

Olivier FOURDAN wrote:

> The speed of a window manager doesn't come from its ability to map windows as quickly as possible, but to respond to user (ie resize/move windows).

I think an important point to note here also is that, when you're
considering application startup, the WM mapping the window is a tiny
tiny fraction of the startup time.  Reading the binary from disk,
loading shared libraries, doing library relocations, then the app doing
whatever it needs before it shows its first window, etc., etc., etc.
will take up a huge majority of the time before you see the window
appear on-screen.

Now, the setup time decrease a bit when you consider an already-running
app opening another window, but my gut feeling here (yay for being
scientific) is that the app creating the window and its children
probably still takes up the bulk of the time.

Optimising the WM's mapping speed here is a waste of time: a classic
case of useles optimisation of a non-critical path; a user wouldn't
likely notice any difference.  I'd rather have the things Olivier
describes: intelligent window placement, fast response to resize and
move, etc.

So: oh yay, E17 maps windows quickly.  Whoopie-do.


Version: GnuPG v1.4.1 (MingW32)


More information about the Xfce4-dev mailing list