Tag browser partitioning & letters

[ownedbycats]

I thought of posting this on the bug tracker as an enhancement request but wanted to make sure it was plausible first.

Currently there's two ways to partition. Either by each individual letter or partitioned.

With first letter, you get a separate partition even if there's only one entry for that letter. With partitioned, you get a fixed amount of entries, but sometimes the first letters get split up.

Would it be very much effort to add a third way that's kind of a mix of the two? e.g. 0-9, A-D, E-H, I-J?. So that it has less partitions then first letter but doesn't split up the entries sharing a first letter.

EDIT: I also noticed this when using first-letter partitioning and I was half-hoping that this might also be a way to partition this by century or decade

[chaley]

Originally Posted by ownedbycats

EDIT: I also noticed this when using first-letter partitioning and I was half-hoping that this might also be a way to partition this by century or decade

Attachment 182700

I thought I understood what you wanted until you made the edit. Are you asking to partition by arbitrary values depending on the category? It seems that you want to sometimes partition by multiple first letters if the number of items in the category is "small", and sometimes partition by something else (e.g., decades or 2 first letters). I think this might end up as specifying DWIM (Do What I Mean).

I might be able to add partitioning first letters, combining first letters to ensure that the number of items in the combination is at least X, giving what you suggested in your fourth paragraph. Your edit suggests doing different combinations depending on the field, which I am not willing to attempt.

If 'Year Published' is a composite column then I think you can solve your problem by making it hierarchical with the first level being the first 3 digits of the year followed by a zero, then '.' then the actual year. Examples: 2010.2015, 2010.2018, 2000.2006.

[ownedbycats]

I think "partition by multiple first letters is the number of items in the category is small" is closest to what I want. I mostly want to keep the entries that share the first letter grouped together instead of splitting them like the Partitioned does.

Year Published is a comp column pulled from the pubdate column. I'll poke around with the template and see if I can figure it out how to set that.

[chaley]

Originally Posted by ownedbycats

Year Published is a comp column pulled from the pubdate column. I'll poke around with the template and see if I can figure it out how to set it hierarchy.

Try making it hierarchical with a template like

program:
	x = format_date(field('pubdate'), 'yyyy');
	strcat(floor(divide(x, 10)), '0.', x)

[chaley]

This template works better.

program:
	p = format_date(raw_field('pubdate'), 'yyyy');
	if p then
		strcat(re(p, '^(...).*', '\1'), '0.', p)
	fi

You probably want to turn off partitioning for this column.

[ownedbycats]

I added an "else 'Unknown'" just before the fi and it works perfectly. Thank you

[chaley]

Originally Posted by ownedbycats

With first letter, you get a separate partition even if there's only one entry for that letter. With partitioned, you get a fixed amount of entries, but sometimes the first letters get split up.

Would it be very much effort to add a third way that's kind of a mix of the two? e.g. 0-9, A-D, E-H, I-J?. So that it has less partitions then first letter but doesn't split up the entries sharing a first letter.

I added an option to tag browser first letter partitioning to combine letters if the number of items under a letter is less than some value you set. If you use this you will end up with single letters if the group of items is large enough, or with (e.g.) "A - F" if letter groups are combined. It is disabled by default (for the moment). It will be in the next release.

Thanks for the suggestion.

[ownedbycats]

Thank you!