Inaccurate counts in section "versions"

Post Reply
bimo
Posts: 113
Joined: 29 Aug 07 22:34
Location: Germany

Inaccurate counts in section "versions"

Post by bimo » 02 Feb 20 19:33

In this section, I find the figures behind each of the defined versions far away from being reliable; is that a specific problem with my installation (and is then there a way to rebuild the figures?) or is it a bug to be reported in Mantis?
Michael

Hert
Posts: 6210
Joined: 13 Sep 03 7:24

Re: Inaccurate counts in section "versions"

Post by Hert » 02 Feb 20 21:47

The counts are fine here. Why do you think they are incorrect?

BTW; they are queried "live" from the database, no precalculations or cached numbers
This is a User-to-User forum which means that users post questions here for other users.
Feature requests, change suggestions, or bugs can be logged in the ticketing system

bimo
Posts: 113
Joined: 29 Aug 07 22:34
Location: Germany

Re: Inaccurate counts in section "versions"

Post by bimo » 02 Feb 20 22:30

Well, the count for version sets is > 40k, that is reliable. I would expect that summing up all of the versions im this section can‘t be less than the count for the sets - but it is by far. I will give you exact figures for a specific version tomorrow.
Michael

Hert
Posts: 6210
Joined: 13 Sep 03 7:24

Re: Inaccurate counts in section "versions"

Post by Hert » 02 Feb 20 22:59

A version doesn't have to be in a placeholder so the number of the placeholders summed up can be less than the count for the sets.
This is a User-to-User forum which means that users post questions here for other users.
Feature requests, change suggestions, or bugs can be logged in the ticketing system

bimo
Posts: 113
Joined: 29 Aug 07 22:34
Location: Germany

Re: Inaccurate counts in section "versions"

Post by bimo » 02 Feb 20 23:33

Mmmh - if a placeholder is defined for a version: isn‘t it then applied to the versioned file by version detection? Can I control the version/placeholder mapping by other means than defining it im the preferences?
Michael

Hert
Posts: 6210
Joined: 13 Sep 03 7:24

Re: Inaccurate counts in section "versions"

Post by Hert » 03 Feb 20 10:00

Best to read through the Versions and Stacks manual first.
Click the ? icon in the left sidebar and select "Quick Manual - Versions and Stacks".
This is a User-to-User forum which means that users post questions here for other users.
Feature requests, change suggestions, or bugs can be logged in the ticketing system

bimo
Posts: 113
Joined: 29 Aug 07 22:34
Location: Germany

Re: Inaccurate counts in section "versions"

Post by bimo » 25 Feb 20 12:52

Sorry that I'm coming back so lately to this topic.

Well, I read the manual but found no hint to when and when not a placeholder is being assigned to a version. As far as I remember, a specific placeholder can only be assigned once in a version set, even if multiple files fullfil a version's condition. The second reason I can imagine is that a placeholder (and the dedicated detection method) have been deleted after the version had been assigned so that the placeholder isn't any longer available in the prefs.

But, honestly, I'm a bit surprised that a section called "versions" is based on the selection criteria "placeholder".

Moreover, I don't understand the huge differences between versions' and placeholders' figures: e.g. I have a version definition for files that have been treated by DxO Photo Lab. The resulting file is marked by the suffix "DxO" and of course the detection method is "file mask". And yes, the first 14 characters are common to both files' name (main and subversion). This is a definition that always worked well since the old days of IDimager... Now what I wasn't aware of are the different figures: 42.496 files are DxO-versions and have reliably been found by version detection - but only 2.521 are shown in the provided section "versions".

I'm at a loss with that situation - is it possible to heal the non-assignment of placeholders? Any other suggestions?
Michael

Hert
Posts: 6210
Joined: 13 Sep 03 7:24

Re: Inaccurate counts in section "versions"

Post by Hert » 25 Feb 20 22:21

Not sure where your statement comes from that versions are calculated based on criteria placeholders. That is incorrect.

An image can have zero, one, or more versions. Together they form a version set. Within a single version set there are banks called placeholders. One of the versions in the set can (not must) occupy a placeholder. The same version can also occupy other placeholders.
This is a User-to-User forum which means that users post questions here for other users.
Feature requests, change suggestions, or bugs can be logged in the ticketing system

fbungarz
Posts: 1559
Joined: 08 Dec 06 5:03
Location: Arizona, USA

Re: Inaccurate counts in section "versions"

Post by fbungarz » 26 Feb 20 19:46

is it possible to heal the non-assignment of placeholders? Any other suggestions?
One of the versions in the set can (not must) occupy a placeholder. The same version can also occupy other placeholders.
Now I am beginning to understand, why assigning placeholders is kind of a mess...
I always assumed "one version = one placeholder", presumably to be assigned automatically via the file naming mask. Now, if the same version can be assigned to two different placeholders and both placeholders have a naming rule how a version matches a particular placeholder - how does that work?

I would have thought the naming rule "forces" the version to be assigned to a placeholder. But reading through this, I am now confused, because apparently the naming rule can then manually be "overruled" by assigning the same version to a different placeholder as well?

I have noticed for quite some time that even though essentially all of my images are part of version sets, following strict naming conventions, making sure the placeholder assignments actually "stick" has always been a struggle. Virtually every time I add a version to the set, it seems that some placeholder assignments get lost and I manually have to force these versions back to the placeholder where they belong.

In my workflow, I shoot JPG+RAW (*.nef), download the pairs into PSu, assigning the JPG as the main version, the *.nef is assigned to my "original" placeholder, I run the DNG converter to generate a second "DNG" version. Then I select the ones I want to edit and via DxO (or in Photoshop) generate JPGs with the naming extension "_print". Back in PSu these get detected via scanning the folders for versions and added as the "print" version". But adding those edited files to a version set for some obscure reason typically trashes the original placeholder assignments and I manually have to re-assign the *.nef again to the "original" placeholder and the *.dng to the DNG placeholder.
I have been getting used to do this as part of my workflow, but never quite understood, why placeholder assignments would simply not "stick". I guess this may in part explain this...

bimo
Posts: 113
Joined: 29 Aug 07 22:34
Location: Germany

Re: Inaccurate counts in section "versions"

Post by bimo » 28 Feb 20 12:31

Well, my original post dealt with versions only. When you, Hert, introduced the term "placeholder" to the discussion, I interpreted that remark as if the assignment of placeholder(s) would be crucial for a version to be displayed in the versions' section.

Now the first and very basic question is: is there a dependency to placeholders if or if not a certain version is being calculated in the versions section? And, perhaps still more basic: what is the aim of this section?
Michael

Hert
Posts: 6210
Joined: 13 Sep 03 7:24

Re: Inaccurate counts in section "versions"

Post by Hert » 28 Feb 20 12:58

Now the first and very basic question is: is there a dependency to placeholders if or if not a certain version is being calculated in the versions section?
I assume you're asking about this:
Versions.png
Versions.png (75.06 KiB) Viewed 3924 times

The red arrow pointing to version sets: this counts the number of Version Sets, and is not the sum of all versions. As explained before: An image can have zero, one, or more versions. If there is at least one versions then it's called a Version Set. In other words; you can have one image with 10 versions which counts for 1 Version Set.

The green arrows pointing to placeholders: this counts the number of images in all version sets that have that place holder occupied. Having a placeholder occupied in a version set is NOT mandatory.

Here's a sample of a catalog with 1 version set. And that one version set has 2 versions. None of the versions are in a placeholder, and so the counts for each placeholder is zero.
Versions2.png
Versions2.png (92.24 KiB) Viewed 3924 times

And here I've stored one of these two versions to every available placeholder.
Versions3.png
Versions3.png (91.81 KiB) Viewed 3924 times

Hope this helps you to understand.

Here's a script that tells you how many versions there are (in my sample above, there is one image with 2 versions and so this script outputs that there are 2 versions):

Code: Select all

  function TotalVersions: Integer;
  var
    ADs: TDBXOMClientDataSet;
  begin
    ADs := PublicCatalog.NewDataSet;
    try
      ADs.CommandText := 'select count(v.GUID) as idCount ' +
                         'from   idImageVersion v ' +
                         'inner join idCatalogItem i on i.GUID = v.MainImageGUID ' +
                         '';
      ADs.OpenSet;
      try
        result := ADs.FieldValue('idCount');

      finally
        ADs.CloseSet;
      end;
    finally
      PublicCatalog.FreeDataSet(ADs);
    end;
  end;

begin
  Say2(
        'Images in catalog: ' + IntToStr(PublicCatalog.TotalImages),
        'Versions in catalog: ' + IntToStr(TotalVersions)
      );
end;
And, perhaps still more basic: what is the aim of this section?
So you can access all images that are stored in a specific placeholder.

Now back to your original question:
Well, the count for version sets is > 40k, that is reliable. I would expect that summing up all of the versions im this section can‘t be less than the count for the sets
You have >40k version sets. And now you should understand that the sum of the versions in the placeholders can be less than the count for the sets.
This is a User-to-User forum which means that users post questions here for other users.
Feature requests, change suggestions, or bugs can be logged in the ticketing system

bimo
Posts: 113
Joined: 29 Aug 07 22:34
Location: Germany

Re: Inaccurate counts in section "versions"

Post by bimo » 28 Feb 20 15:53

Dear Hert, thank you so much for your very detailed explanation that shows my complete misunderstanding of the section: I expected every single version of a version set instead of the placeholders to be counted in this section. (Still, I believe that "Placeholder" would be the more precise title for this section...).

I obviously misunderstood already the versioning part in the prefs: I not only defined the main version detection method by a file mask but also defined each and every version I aimed to detect on the following lines; I thought this to be mandatory for the process of version detection. And the placeholder? Just an attribute to the specific version...

Despite my misunderstanding I'm for years successfully working with versioning which is at least one of the most important features (if not the most important one) PSU has to offer. BTW I solved my requirement to count each and every version in all version sets by a new category "Versions" and ran the sript "Macro Command to Catalog Label" from repository: the filename's last three characters make up a version in my system.

Thanks again for your clarifying answer!
Michael

Post Reply