Re: counting downloads
FYI, Vadim is already providing stats on some (non incubating) projects: http://people.apache.org/~vgritsenko/stats/projects/index.html I don't know if it's easy for him to add incubating projects, and how he deals with mirrors. In a sense even if the download counter is not accurate, I agree with eelco to say that the trend can still be interesting. Xavier On 5/2/07, Marshall Schor [EMAIL PROTECTED] wrote: We're curious to see how many downloads we're getting, perhaps sorted by ip number or who's downloading (I realize that would need be volunteered information). Do other projects have a good way to track this? I know we could pull the logs for the p.a.o webserver and grep through them looking for things, but I'm wondering if there's something we can put on our download page that users would click on that would count things? -Marshall Schor - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED] -- Xavier Hanin - Independent Java Consultant Manage your dependencies with Ivy! http://incubator.apache.org/ivy/ - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
Re: counting downloads
On 5/3/07, Daniel Kulp [EMAIL PROTECTED] wrote: Noel, On Thursday 03 May 2007 03:59, Noel J. Bergman wrote: Except you forget one detail: incubator artifacts don't go to the mirror system. And that's changing/changed. Can I ask when?The vote that was called in mid march [1] never had a result posted. The discussion [2] that started that vote also didn't seem to reach any type of concensus. we've now had a clear call from infra (incubator works within broader apache policy) i've done this kind of infrastructure work before so i'll pull together a plan in JIRA and run it past people on this list and infra I guess my major concern right now is the CXF/Wicket releases that have gone out in the lasts couple days. Did we do the right thing just putting them in /www/people.apache.org/dist/incubator or should they have gone to the mirrors? There isn't a /www/www.apache.org/dist/incubator directory. Anyway, I'd just like to make sure we didn't do something wrong due to a policy change I missed. (I could easily have missed it, email overload and all) don't worry - IIRC the policy hasn't changed yet IMHO this is something that needs to be fixed by the IPMC. it's going to take a little while to pull everything together. we have some scripts in jakarta that generate the download pages from meta-data and that's the direction i think we need to move in. (they also have some cool RSS/news stuff for releases is cool) i'd rather have CXF and wicket in the old places and have the time to do this properly than delay the releases or rush the move - robert - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
Re: counting downloads
On 5/3/07, Danny Angus [EMAIL PROTECTED] wrote: On 5/2/07, Garrett Rooney [EMAIL PROTECTED] wrote: Do other projects have a good way to track this? I know we could pull the logs for the p.a.o webserver and grep through them looking for things, but I'm wondering if there's something we can put on our download page that users would click on that would count things? James uses google analytics. We currently count the hits to the download cgi page, which is probably over generous, but you can put google javascript on the links to the files themselves, we just haven't done it. i'm going to need to create download pages for the incubator. this isn't one of my personal itches and i would need to research the analytics-foo. but if someone wants to create a JIRA with the required javascripts, i'll take a look at including it. any volunteers? - robert - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
Re: counting downloads
Thanks for the clarification Robert. I'm glad we didn't mess something up. :-) I'll definitely keep an eye on the various threads for information about the changes. It all sounds very good to me. :-) Dan On Sunday 06 May 2007 06:55, robert burrell donkin wrote: On 5/3/07, Daniel Kulp [EMAIL PROTECTED] wrote: Noel, On Thursday 03 May 2007 03:59, Noel J. Bergman wrote: Except you forget one detail: incubator artifacts don't go to the mirror system. And that's changing/changed. Can I ask when?The vote that was called in mid march [1] never had a result posted. The discussion [2] that started that vote also didn't seem to reach any type of concensus. we've now had a clear call from infra (incubator works within broader apache policy) i've done this kind of infrastructure work before so i'll pull together a plan in JIRA and run it past people on this list and infra I guess my major concern right now is the CXF/Wicket releases that have gone out in the lasts couple days. Did we do the right thing just putting them in /www/people.apache.org/dist/incubator or should they have gone to the mirrors? There isn't a /www/www.apache.org/dist/incubator directory. Anyway, I'd just like to make sure we didn't do something wrong due to a policy change I missed. (I could easily have missed it, email overload and all) don't worry - IIRC the policy hasn't changed yet IMHO this is something that needs to be fixed by the IPMC. it's going to take a little while to pull everything together. we have some scripts in jakarta that generate the download pages from meta-data and that's the direction i think we need to move in. (they also have some cool RSS/news stuff for releases is cool) i'd rather have CXF and wicket in the old places and have the time to do this properly than delay the releases or rush the move - robert - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED] -- J. Daniel Kulp Principal Engineer IONA P: 781-902-8727C: 508-380-7194 [EMAIL PROTECTED] http://www.dankulp.com/blog - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
Re: counting downloads
Noel J. Bergman wrote: Garrett Rooney wrote: The mirror system makes this an essentially unsolvable problem. And in addition to all of your other valid points, there is the problem of having a a farm of caching proxies on the ISP side of the net, leading to the seemingly odd case that the more popular a download, the less likely you are to get a good count, because the more likely it is to be served from the caching proxies. --- Noel One confusion I have: If we count clicks on the download link, it seems that even if that link led to a mirror page, it would count pretty accurately (except of course if a person clicked to download, and then didn't bother going through with it). I guess it would also miss the case where people somehow found their way to a download mirror *without* going through the project download page. Did I miss something, or would this give a reasonable estimate of downloads? -Marshall - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
Re: counting downloads
On 5/5/07, Marshall Schor [EMAIL PROTECTED] wrote: One confusion I have: If we count clicks on the download link, it seems that even if that link led to a mirror page, it would count pretty accurately (except of course if a person clicked to download, and then didn't bother going through with it). I guess it would also miss the case where people somehow found their way to a download mirror *without* going through the project download page. Did I miss something, or would this give a reasonable estimate of downloads? Unless they don't download it via your link, or they download more than once (getting copies on multiple machines?), or any number of other things that can throw your numbers off. It's a losing battle for statistics that IMO aren't very useful anyway. All download counts are good for is ego stroking, there are better ways to spend time and energy. -garrett - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
Re: counting downloads
Unless they don't download it via your link, or they download more than once (getting copies on multiple machines?), or any number of other things that can throw your numbers off. It's a losing battle for statistics that IMO aren't very useful anyway. All download counts are good for is ego stroking, there are better ways to spend time and energy. I agree with most of the comments here. But one thing download statistic imho *are* useful for is measuring progress. Even if the absolute numbers aren't useful, and trends are not reliable, it definitively gives an idea of whether a project is gaining attention or not. Another thing is that even if statistics aren't reliable, people still often want to have them. Matt Raible's talk on ApacheCon listed job counts per framework on Dice for instance. Completely useless, but people seem to love to hear it, even at Apache. Or there is my publisher asking me for download statistics of the project I'm involved in and using Amazon's ranking of a competing book to 'determine' the market potential for the topic. Imho, it's enough to say that we/ Apache doesn't want to spend energy on something that is technically hard or impossible and that in the opinion of the people who have to support it serves little value. Eelco - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
RE: counting downloads
Except you forget one detail: incubator artifacts don't go to the mirror system. And that's changing/changed. --- Noel - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
Re: counting downloads
Noel, On Thursday 03 May 2007 03:59, Noel J. Bergman wrote: Except you forget one detail: incubator artifacts don't go to the mirror system. And that's changing/changed. Can I ask when?The vote that was called in mid march [1] never had a result posted. The discussion [2] that started that vote also didn't seem to reach any type of concensus. I guess my major concern right now is the CXF/Wicket releases that have gone out in the lasts couple days. Did we do the right thing just putting them in /www/people.apache.org/dist/incubator or should they have gone to the mirrors? There isn't a /www/www.apache.org/dist/incubator directory. Anyway, I'd just like to make sure we didn't do something wrong due to a policy change I missed. (I could easily have missed it, email overload and all) [1]http://mail-archives.apache.org/mod_mbox/incubator-general/200703.mbox/31cc37360703151823u2099483dn95aeaed8c6752245%40mail.gmail.com [2]http://mail-archives.apache.org/mod_mbox/incubator-general/200703.mbox/31cc37360703140920h6a601b57gbb86ff0cd5e8049b%40mail.gmail.com -- J. Daniel Kulp Principal Engineer IONA P: 781-902-8727C: 508-380-7194 [EMAIL PROTECTED] http://www.dankulp.com/blog - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
Re: counting downloads
On 5/2/07, Garrett Rooney [EMAIL PROTECTED] wrote: Do other projects have a good way to track this? I know we could pull the logs for the p.a.o webserver and grep through them looking for things, but I'm wondering if there's something we can put on our download page that users would click on that would count things? James uses google analytics. We currently count the hits to the download cgi page, which is probably over generous, but you can put google javascript on the links to the files themselves, we just haven't done it. d. - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
Re: counting downloads
Yes, I'd be interested in this information for the Apache Tuscany downloads as well... On 5/2/07, Marshall Schor [EMAIL PROTECTED] wrote: We're curious to see how many downloads we're getting, perhaps sorted by ip number or who's downloading (I realize that would need be volunteered information). Do other projects have a good way to track this? I know we could pull the logs for the p.a.o webserver and grep through them looking for things, but I'm wondering if there's something we can put on our download page that users would click on that would count things? -Marshall Schor - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED] -- Luciano Resende http://people.apache.org/~lresende
Re: counting downloads
On 5/2/07, Marshall Schor [EMAIL PROTECTED] wrote: We're curious to see how many downloads we're getting, perhaps sorted by ip number or who's downloading (I realize that would need be volunteered information). Do other projects have a good way to track this? I know we could pull the logs for the p.a.o webserver and grep through them looking for things, but I'm wondering if there's something we can put on our download page that users would click on that would count things? The mirror system makes this an essentially unsolvable problem. Not to mention the fact that you have no clue if people are actually getting the code from one of our mirrors at all, they could get it from a linux distribution, or any number of other repackagers. On top of that, you can't correlate back to an individual person, so you don't know if you're getting one guy downloading it more than once because he's accidentally deleted it... Get used to the idea that counting downloads is an inherently poor way to judge if your software is being used. More interesting metrics might be numbers of posts on mailing lists, bugs filed, etc. -garrett - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
RE: counting downloads
Garrett Rooney wrote: The mirror system makes this an essentially unsolvable problem. And in addition to all of your other valid points, there is the problem of having a a farm of caching proxies on the ISP side of the net, leading to the seemingly odd case that the more popular a download, the less likely you are to get a good count, because the more likely it is to be served from the caching proxies. --- Noel - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
Re: counting downloads
On Wednesday 02 May 2007 18:43, Garrett Rooney wrote: The mirror system makes this an essentially unsolvable problem. Not to mention the fact that you have no clue if people are actually getting the code from one of our mirrors at all, they could get it from a linux distribution, or any number of other repackagers. On top of that, you can't correlate back to an individual person, so you don't know if you're getting one guy downloading it more than once because he's accidentally deleted it... Except you forget one detail: incubator artifacts don't go to the mirror system. (Which, again, I think is silly. Part of being in the incubator is to learn about apache processes. One of those should be the mirror system.) To answer the question: all the logs are available on people.apache.org:/x1/logarchive/people You can parse them an pull out any information you want. Just keep in mind that after you graduate, all warnings about the mirrors do apply, so don't get used to having the download information. -- J. Daniel Kulp Principal Engineer IONA P: 781-902-8727C: 508-380-7194 [EMAIL PROTECTED] http://www.dankulp.com/blog - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
Re: counting downloads
Thanks Dan, indeed, as Tuscany is still under incubation, mirrors and other things won't apply. Did anybody ever created a script to parse the logs and provide totals ? On 5/2/07, Daniel Kulp [EMAIL PROTECTED] wrote: On Wednesday 02 May 2007 18:43, Garrett Rooney wrote: The mirror system makes this an essentially unsolvable problem. Not to mention the fact that you have no clue if people are actually getting the code from one of our mirrors at all, they could get it from a linux distribution, or any number of other repackagers. On top of that, you can't correlate back to an individual person, so you don't know if you're getting one guy downloading it more than once because he's accidentally deleted it... Except you forget one detail: incubator artifacts don't go to the mirror system. (Which, again, I think is silly. Part of being in the incubator is to learn about apache processes. One of those should be the mirror system.) To answer the question: all the logs are available on people.apache.org:/x1/logarchive/people You can parse them an pull out any information you want. Just keep in mind that after you graduate, all warnings about the mirrors do apply, so don't get used to having the download information. -- J. Daniel Kulp Principal Engineer IONA P: 781-902-8727C: 508-380-7194 [EMAIL PROTECTED] http://www.dankulp.com/blog - To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED] -- Luciano Resende http://people.apache.org/~lresende
