Appending Tags To Existing QVDs

[cotiso_hanganu]

Hi there,

I am trying to improve the way QVD are used in some of our data sets. I am using tables (&fields, maybe later ) tags to add metadata to resident table and afterwards store them in QVD files. Reading afterwards only the tags from the QVD can make a huge difference when big (or remote) QVD are in the game.

Still, I wish I could add some times some additional tabel tags to the QVD without having to load resident again the whole table , only to add/modify the table tags and afterwards store it again (in same/another location).

Kind of appending the tags metadata in the XML header only .

Any idea if this is possible some how ?

Or maybe could this become relevant for more within the community so that we can push it to ideation ?

@stevedark

@rwunderlich

@hic

@barryharmsen

[stevedark]

Nice idea. Not sure of a way of achieving this though.

I would have thought that most of the QVDs would be written to every now and again anyway, so there would be possibility to add tags at that point? If you have archive QVDs, for older periods for example, then it would just be a one off to append the tags, so could be scheduled over a weekend?

Another approach is to have a separate metadata table somewhere, which could be seeded from QVD metadata but then be available to be updated with additional tags. This could also contain history of size of QVDs etc.. The downside of this approach though is that the metadata is not contained within the QVD itself.

Steve

[marcus_sommer]

Without a reload it's probably not possible to add or change tags or comments. You may open the qvd within an editor and then manipulating the xml-header but there are serious risks to get a corrupted file after the saving and in regard to the needed efforts it's rather not a benefit ...

For just a few extra information you may include them within the filename like:

BasisName_Detail_YYYYMM#Tag1#Tag2.qvd

and all load-logig ignored the parameter-area. I use a similar logic often to add file/table-information which are later extracted within filelist-loops but not for tags or comments.

IMO you should follow the suggestion from @stevedark to maintain the needed meta-information within a separate and maybe centralized source - which might be also used as mapping source for all tags/comments in the final reports and/or as an extra development/governance report.

- Marcus