[pdftex] OT: Linux AR 7.0 released: another comment
James Quirk
jjq at galcit.caltech.edu
Sat Apr 30 17:19:52 CEST 2005
> If they did it right, all versions would work identically, have the
> same GUI appearance, and be released on the same day. To support this
> model, I told them that in xpdf, a GPLed PDF viewer for the X11 Window
> System, only about 10% of the code is devoted to the window interface;
> I expect that the figures for Acrobat should be similar.
I couldn't agree more with the software strategy that Nelson puts forward.
However I'm afraid there are far more serious failings of AR 7.0 than
the GUI problems discussed above. For instance, here is a
performance comparison, between AR 5.1 and AR 7.0, for the time
it takes to turn a collection of widget annotations on and off:
Number | AR 5.1 | AR 7.0
of widgets | On | On Off
----------------------------------------------------------------
10 | 0.002 | 0.002 0.002
50 | 0.02 | 0.08 0.10
100 | 0.05 | 0.29 0.42
500 | 0.96 | 8.34 24.88
1000 | 3.86 | 40.61 175.81
1500 | 9.25 | 113.57 571.77
2000 | 17.81 | 235.58 1344.11
1. All timings are in seconds
2. Timings were measured on an IBM T42 (1.6 MHz Pentium M)
running SuSE 9.1 .
3. For AR 5.1 the time to turn the widgets off is the same
as for turning them on.
4. The test document use JavaScript; e-mail me if you would
like a copy.
The degradation in performance is alarming: AR 7.0 is seventy five times
slower than AR 5.1 in turning off 2,000 widgets. This would suggest that
Adobe's software testing does not run to checking how well the PD format
scales. But this would be in keeping with the picture that Nelson paints.
James
p.s. Before dismissing the present perfomance test as artificial,
you might care to look at the documents:
http://www.amrita-ebook.org/doc/asme/cfd-policy
http://www.amrita-ebook.org/doc/amr2003
both of which employ 100's of widget annotations to good effect.
More information about the pdftex
mailing list