Anca Paula Luca a écrit : > Guillaume Lerouge wrote: > >> Hi, >> >> On Wed, May 6, 2009 at 6:31 PM, Ludovic Dubost <[email protected]> wrote: >> >> >>> Thomas Mortagne a écrit : >>> >>>> On Wed, May 6, 2009 at 17:09, Ludovic Dubost <[email protected]> wrote: >>>> >>>> >>>>> I'm -1 on this. >>>>> >>>>> This is like saying we should put the fully qualified name when >>>>> >>>>> - linking to an anchor in your page >>>>> - linking to a page in the same space >>>>> >>>>> >>>> This example as nothing to do with what i suggest since this is user >>>> input and not technical generated datas. I'm speaking about internal >>>> generated datas, datas you don't find in the export, the link table is >>>> not exported etc... >>>> >>>> >>> I tend to think that what I say is true for any data. >>> As Anca points out a mysql backup of the backlinks data with absolute >>> links would not work if you recover it under another db name. >>> >> There's maybe something I don't understand here, but isn't it better to have >> more data and run a filter to remove unnecessary data in specific cases if >> needed (such as the MySQL backup scenario decribed above) than to have too >> little data and to get stuck when trying to identify which wiki a given page >> belongs to? Would writing such a filter be complex? >> Whatever has to treat data if you import/export is going to be a pain.
In this specific case we would also need a migration since previously stored backlinks data was NOT including the wiki name. And code based on the fact that it's not there will fail. >> That is, it's easier to have too much data and to filter out the noise than >> the other way round, isn't it? >> > > Too less data is bad because, well, you don't have the data. > Too much data is also wrong because, when it gets redundant (as it would be > the > case here), updating can become a pain, cause inconsistencies, bring you back > to > the first situation when you didn't have data (but this time only because it > doesn't match and you don't know what's the real data). > By default, redundancy is to be avoided, unless you know exactly what you're > doing and it really improves performance. > exactly. We already know in which wiki we are. No need to repeat it everywhere. If you do you'll have to run script to modify it if you change the wiki name. > Anca > > >> Guillaume >> >> >> >>> Ludovic >>> >>>>> We should always store information as "relative". If a document is in >>>>> the same wiki then you should not specify the wiki. >>>>> If it is in the same space you should not specify the space. >>>>> >>>>> Imaging you want to export/import a wiki that was in a farm in a >>>>> database names 'wiki1' and you import the pages in a farm where the >>>>> database is names 'wiki2'. Then your import process will have to run >>>>> conversions, or you wiki will be fully broken. >>>>> >>>>> I'm for generalizing "relative" storage if there are places where we >>>>> hard code the wiki name or the space name. >>>>> >>>>> Ludovic >>>>> >>>>> Thomas Mortagne a écrit : >>>>> >>>>> >>>>>> Hi devs, >>>>>> >>>>>> The real full qualified name of a document is wiki:Space.Page. I think >>>>>> we sould always use only this form for technical purposes like storing >>>>>> in database or old API which does not support DocumentName. >>>>>> >>>>>> pros: >>>>>> - don't need to take care of the context, the information is always >>>>>> >>> right. >>> >>>>>> - if everything is stored in only one form it's way easier to compare >>>>>> things and do requests >>>>>> >>>>>> cons: >>>>>> - old code which does not support this document name form will fail. >>>>>> This could hardly be a good arguments since this form is a officially >>>>>> valid document name but i can't see anything else. >>>>>> >>>>>> Among others, the last use case which make me send this proposal is >>>>>> http://jira.xwiki.org/jira/browse/XWIKI-3754. IMO getLinkedPages >>>>>> return and the link table should only contains fully qualified links, >>>>>> especially since theses lists are supposed to contains uniques >>>>>> documents and we need to know the 3 parts of the document name to make >>>>>> sure of that . Note that a document could contain a link to a document >>>>>> in another wiki or even a link to the same document but written in its >>>>>> full form so a code which support only local form is wrong anyway. >>>>>> >>>>>> WDYT ? >>>>>> >>>>>> Here is my +1. The pros are basically the same that make us introduce >>>>>> DocumentName to not manipulate String anymore in code. >>>>>> >>>>>> -- >>>>>> Thomas Mortagne >>>>>> _______________________________________________ >>>>>> devs mailing list >>>>>> [email protected] >>>>>> http://lists.xwiki.org/mailman/listinfo/devs >>>>>> >>>>>> >>>>>> >>>>>> >>>>> -- >>>>> Ludovic Dubost >>>>> Blog: http://blog.ludovic.org/ >>>>> XWiki: http://www.xwiki.com >>>>> Skype: ldubost GTalk: ldubost >>>>> >>>>> _______________________________________________ >>>>> devs mailing list >>>>> [email protected] >>>>> http://lists.xwiki.org/mailman/listinfo/devs >>>>> >>>>> >>>>> >>>> >>>> >>> -- >>> Ludovic Dubost >>> Blog: http://blog.ludovic.org/ >>> XWiki: http://www.xwiki.com >>> Skype: ldubost GTalk: ldubost >>> >>> _______________________________________________ >>> devs mailing list >>> [email protected] >>> http://lists.xwiki.org/mailman/listinfo/devs >>> >>> >> >> > _______________________________________________ > devs mailing list > [email protected] > http://lists.xwiki.org/mailman/listinfo/devs > -- Ludovic Dubost Blog: http://blog.ludovic.org/ XWiki: http://www.xwiki.com Skype: ldubost GTalk: ldubost _______________________________________________ devs mailing list [email protected] http://lists.xwiki.org/mailman/listinfo/devs
