Pages and page-titles confusion after upgrade >=0.5 - fix please

hi piwik-developers

i opened up this problem-report already in general discussion (301 Moved Permanently) months ago but i dont think someone really got into it and sees the problem that arises! generally i see “pages” that are now “page-titles” and “urls” that now also are “page-titles”. but urls will never be page-titles. such a “conversion” is not correct!

please check your demo piwik setup which shows exactly the problem also. you did upgrade demo on 08.12.2009 so data before and after are presented differently. even worse if you look at time-frames with the upgrade in between.

detailed explanation:

a) old “urls” now in wrong category:

old “urls” that were visible under “pages” before upgrade to 0.5 and now are not visible under “pages” anymore (imho this is wrong!):
piwik.org/demo/index.php?module=CoreHome&action=index&idSite=1&period=day&date=2009-12-07#module=Actions&action=getPageUrls&idSite=1&period=day&date=2009-12-07

old “urls” that were visible under “pages” are now visible under “page-titles” after upgrade to 0.5 (imho this is wrong!):
piwik.org/demo/index.php?module=CoreHome&action=index&idSite=1&period=day&date=2009-12-07#module=Actions&action=getPageTitles&idSite=1&period=day&date=2009-12-07

b) new urls - correct

new “urls” after upgrade are visible under “pages” (this is correct!):
piwik.org/demo/index.php?module=CoreHome&action=index&idSite=1&period=day&date=2009-12-08#module=Actions&action=getPageUrls&idSite=1&period=day&date=2009-12-08

new “page-titles” after upgrade are visible under “page-titles” (this is correct!):
piwik.org/demo/index.php?module=CoreHome&action=index&idSite=1&period=day&date=2009-12-08#module=Actions&action=getPageTitles&idSite=1&period=day&date=2009-12-08

c) timeperiod with upgrade in between - wrong values

if one views a period where inside the upgrade to 0.5 was made, one sees a mix of “page-titles” and “urls” under “page-titles” (this is wrong!):
piwik.org/demo/index.php?module=CoreHome&action=index&idSite=1&period=week&date=2009-12-08#module=Actions&action=getPageTitles&idSite=1&period=week&date=2009-12-08

d) proposed solution

  • upgrade does not change anything on old “pages”. just keep them there where they are: under “pages”! as you can not decide to make “page-titles” out of “urls”.
  • upgrade does ask if “pages” should also be visible under “page-titles”. (but default is no)

e) our situation
we’re running piwik in a large scale environemnt with some thousand users (several 10’000 pages with around 3’000’000 page views a day). we have integrated user and page creation in piwik completely in our backend. with the problem described above we can not upgrade to piwik >= 0.5 as it breaks up stats (as explained) and we’re not able to explain this to all our customers. so this is a real major stopper for us. it would be very pitty to stop using piwik already, as it really looks promising and we would like to use it even further for application-internal-stats with lots of custom plugins.

if you need further information or want to get in contact with me. please write me directly to the email-add which i registered for this forum-account. i’m open for any kind of solution for this (custom patch that works, sponsored development, …) as i’m not the only person with same problem, but maybe the one with the biggest impact on business.

thanx and greets.ivo

i did some tests and i’m not sure if i found the solution to change old data to work correct in newer versions:

it looks like in piwik_log_action, field “type” changing could do the trick:

  • old page: type=1
  • new page: type=4

but i see also new type=1 which is an url. but i’m not sure if this is the referer. which might be type=2 in old versions? or is it a new value?

if this is correct i just need to remove alle processed piwik_archive_* tables. change inside piwik_log_action the type=1 to 4 . right?
(maybe i need to change also some oder types…)

and then i can recalculate everything new with the upgraded codeversion… right?

if i find more time to investigate i’m starting a test on our develpment system. but it would be very cool if you could help me here now with dokumentation about those “state”-field meanings in old an new versions. i think this is not to much work for you and helps us alot, and last but not least all our several thousand users will be happy about the upgrade…