How many users of mageia?

[juergen_harms]

Are there any estimates on the number of people using / having tried Mageia? Would be interesting to have an idea - just at the end of the RC period, and a couple of weeks after the offcial release has come out.

I realise that the download count does not say much about the effective use, but maybe there are cleverer way to get at least a wild guess (for instance number of accesses to fetch updates during a given period)

[Germ]

Yea, that would be nice to know. I don't have any idea.

[obgr_seneca]

I think, that's technically impossible, for you have no way of finding out, how many people downloaded what from which server, and I think we won't get the larger mirrors to do something like that for us...

[rda]

Well, at least we can try to have a rough estimate - in the end, it's more the variation and the scale, rather than the exact value, that is significant.

We have several potential sources of information that we can refine, if we all agree on these - that involves from developers to sysadmin and mirror admins.

1. we can count the number of download requests from the Web site - if we take as a habit to always route people to download from http://mageia.org/downloads/ (and not to go in their FTP mirror instead, to which they would likely be redirected anyway) at some point, if should be somehow representative of most downloads. That says nothing on actual installation and usage rate however.

2. we could suggest mirror admins (that are ok with it) to provide download log (or give them a script to parse/compute their logs) so we can get the request numbers; not all mirrors will provide this, but that should not prevent from offering to do it; even if we have only a few mirrors in this game, we can correlate their numbers with above 1) numbers.

3. we can count requests from mgaapplet to the releases.mageia.org website; at some point, this can give a representative activity of online systems; that have a somehow accurate number would need some work about the refresh/update frequency to this releases server.

4. we can count requests to the mirrors list API on mirrors.mageia.org; but same as above, it needs some work about the frequency to get a somehow acceptable estimate.

5. other ideas?

All in all, this will need 2/3 people to gather and set this up in a working, consistent, approved manner and to publish this in our public dashboard. This relates to the privacy policy as well as we have to make obvious that no identifying info is/will be used in these processes and that the compiled results are to be public.

And, again, this won't provide a very accurate picture, but a rough estimate/scale, still interesting enough.

[wobo]

1. We will not be able to direct users to download from the Mageia download page. At least not in a reasonable percentage, there are too many local pages, forum posts, etc. which point at local mirrors.

2. Asking mirror maintainers is a good idea, already asked by Anne in early April. She promised to send me a script for our server but I guess, she was distracted by daily trade and a loaded desktop

[juergen_harms]

I did not ask a technical question - sorry if my mentioning that I realise this is not evident lead to misunderstandings.

I really think that such figures - even if extremely inaccurate - are essential, precisely in the current phase when articles will be written about the new-comer, and when Mageia has interest to demonstrate that it has already now a weight that counts (which is beyond inaccurate - an optimistic guess).

[rda]

1. We will not be able to direct users to download from the Mageia download page. At least not in a reasonable percentage, there are too many local pages, forum posts, etc. which point at local mirrors.

Well, that may really be a case of a usage to promote as well (not to impose); I think most people will use the canonical URL provided; but of course, that will only give a sense of scale, not of absolute value.

2. Asking mirror maintainers is a good idea, already asked by Anne in early April. She promised to send me a script for our server but I guess, she was distracted by daily trade and a loaded desktop

Ha. We're all under the water :-p Actually, I'm a bit working on that (for the full project dashboard) but am quite unsure how to proceed (first a quick, dirty working hack; or something more toughtful). I guess one of the first part is:
- getting a sense of all log formats we can gather,
- see how to normalize all of them
- and may be stripping what may be problematic before being aggregated into a central place where all stats could be computed from (such as replacing IP addresses with Country/Location info only)

I did not ask a technical question

Yes, but this leads to that. Because in order to provide a reference number (how inaccurate it could be) we need to provide as well the method to get to that number - and see how it evolves with time. So far, the current estimate I can give of downloads triggered from www only is about 31421 since beta 1:
- 12757 for beta1
- 10994 for beta2
- 7597 for RC

(so it does not count: direct downloads from other mirrors/platforms, neither regular system updates between each release, neither alpha ones). Method used: parsing through www access log and matching on the download request URL with some grep, gawk, sort and uniq. Hopefully we will make this way more lovely to read in coming weeks

[juergen_harms]

I guess these figures are iso-downloads + update downloads.

A milkmaids fingercount is still better than nothing at all:

• from 30.4 to 27.5 I launched 37 downloads for fetching update packages (my stats before that date got wiped out when I installed Beta2)
• this period corresponds quite roughly to the period for Beta2 plus RC1, where the count of downloads at Mageia is about 18000
• assuming that my download rate is that of a typical pre-release user: each active user launching downloads from Mageia is assumed to have done 37 downloads for updates and 2 for iso downloading
• that makes 18000 / 39, i.e 460 active users that did downloads from the mageia site
• given the 2 : 39 ratio between Iso and Update downloads, other methods to get the initial image can be considered negligeable

I have no idea about the ratio between downloads at Mageia and downloads from other mirrors - if you can get hold of these figure, you could get a correction factor (btw. my downloads were from the Erlangen and the Geneva mirrors).

That does not sound far off - counting other mirrors, that looks like substantially more than 500 users, quite impressing (unless my estimation has an important flaw)

The advantage of this way of guessing is, that the predominant influence comes from downloads for updates - i.e. signifies active users. Originally I had thought that some counter of install-accept hits would be more precise - it probably is not, since it would include all the "tourists" just making an initial install and than giving up, and multiple temptatives to get a single installation going. The critical figure is the downloads-per-user factor: doing 1.something downloads per day during RC sounds like a reasonable figure, but ...

PS: all at sudden I have a doubt: was my end-of-april reference the launch of Beta1 or of Beta2? if it was Beta1, my figure of 18000 should be corrected - but anyhow, it is necessary to check whether this makes sense

[rda]

I guess these figures are iso-downloads + update downloads.

No, these are only for ISO downloads (and having no way to aggregate logs from mirrors yet, we have no idea of the rate of updates downloads).