It seems my previous post was confusing. The point I was trying to illustrate was that we have varying percentages of downloads coming from IE users that seem heavily skewed depending on whether the download is of, for example, the el-GR localization -- where 86% of the people downloading the Greek localization did so using IE - or the en-GB where only 45% of the downloads were performed using IE.
Another way to put this would be to say that, based on my rough analysis, people seeking the en-GB localization are mostly downloading with non-IE browsers. I don't have a breakdown of that but it's probably mostly older Gecko-based browsers. People seeking the Greek localization are mostly downloading from an IE client and so I'm guessing that before 1.0 Gecko-based clients didn't have a lot of penetration into the Greek market.
Is this more clear? The reason I even bring this up is because I hear a lot of people saying things like "yeah, you may have millions of 1.0 downloads, but those are all probably just people upgrading from the Preview Release." So this was a response of sorts trying to point out that people using IE to download Firefox 1.0 probably aren't PR users (who would presumably be using the PR to download 1.0 rather than IE.)
My first pass through the data suggested that more than 53% of the Windows Firefox 1.0 dowloads were being performed by IE users. Out of curiosity, I looked a bit closer and saw that there was a fairly wide spread, with IE users dominating a few of the localized builds, and non-IE downloaders having a strong majority in other localizations.
Hope this clears things up a bit more. Let me know if it doesn't.
Posted by asa at November 21, 2004 09:23 AMDownload statistics are pretty useless really, considering that one exe install downloaded with IE could be deployed on 50,000 machines. While it's a nice comparison, we shouldn't get to hung up on it.
We should instead focus on the appalling fact that there is no MSI for Firefox yet, this is a disgrace and should of def. blocked Firefox 1.0.
Posted by: Martin Alderson on November 21, 2004 09:58 AMAsa, can you also publish download statistics for the localized builds, please! Thank you!
--Thomas
Posted by: Thomas Kaschwig on November 21, 2004 10:06 AMMartin, there's nothing to be gained by holding the release of 1.0 until an MSI is available. It's still just coming "ASAP" (see also https://bugzilla.mozilla.org/show_bug.cgi?id=231062)
Posted by: Greg K Nicholson on November 21, 2004 10:40 AMHi everybody,
What really counts is browser usage. One copy of Firefox might be deployed on a score of machines, or dumped into the trash. Many people will get the fox from the CD-ROMs packaged with computer magazines. Some Linux distros already deliver Firefox to the customer.
So my question is: Do we have reliable browser detection code? Is there a campaign underway to get said code onto as many servers as possible?
Posted by: adaxl on November 21, 2004 11:00 AMWhat do you mean by 'browser detection code'?
Browsers already send out an ID giving information about the browser and operating platform (i.e. Windows, Linux, etc.). Servers log this, and that's where we get statistics such as who's using what browser to download the thing. It's not just downloads, either- or rather, it is, but every page view is in reality a bunch of downloads of images and HTML files, so it doesn't make a difference.
So, what is there to gain from "browser detection code?" Or am I completely misunderstanding your meaning?
Posted by: nentuaby on November 21, 2004 02:43 PM@ nentuaby: Yes, we all know there is good browser detection code out there, but there are still too many pages that use bad code that does not detect right, or lumps all Gecko browsers into one category, etc.
Posted by: adaxl on November 21, 2004 08:04 PMOneStat.com reports a 4.58% share of Firefox across their 50,000 sites in 100 countries. http://www.onestat.com/html/aboutus_pressbox34.html
Posted by: William on November 22, 2004 08:08 AM