> I'll retest sensors-detect tonight and let's take another > day for testing and reviewing the docs. Sounds OK. I'll test a fresh checkout on my four systems (sensors-detect, drivers and sensors) this afternoon. > We can give Phil the OK on Wednesday and he can release late in the > week.(Phil likes releasing late in the week to avoid weekday load on > his server) That's OK for me. BTW I guess that the high load is caused by the Freshmeat announcement, not by the release itself, so the release day may be chosen freely I suppose. If load is an issue, there are two major things that could be done to limit it: 1* Use bzip2 instead of gzip to compress the release files. It should reduce the network use by 25%. We can make both available if you think there are still persons who can't handle bzip2-compressed files. In this case, I think we can expect a 10% network use reduction. 2* Clean our HTML code. A simple HTML cleanup as the one I did for the New Drivers page can reduce the page size by 2 to 3%. Using Style sheets would probably reduce by much more than that, maybe up to 30 or 40%. Also, we could stop using server-side includes and build the pages using a simple Makefile. This would probably be less CPU-power consuming for the web server. We could also use PNG instead of GIF for our pictures (I'd expect a 20% size reduction). BTW, 5 of our pages have this include: <!--#include virtual="test.cgi" --> It seems to be a counter, but it doesn't work (displays: "0 hits since July 23, 1998"). It's pretty lame isn't it? ;) I think we should either fix the CGI (and rename it to something better than test.cgi) or remove the includes (and delete the script). The question is, do we need a counter? -- Jean Delvare http://www.ensicaen.ismra.fr/~delvare/