Template talk:Irregular category series navigation

Usage suggestion
I was glad to see this template, but when it is used directly on each linked page of a series, this does not minimise future maintenance, as new links need to be added/renamed on each linked page. For each series such as Category:Members of the Scottish Parliament 2016–2021 & siblings, it would be better to build a specific template for that series, e.g. template:Members of the Scottish Parliament by term, using this template. – Fayenatic  L ondon 09:01, 10 May 2021 (UTC)
 * Thanks for your thoughts, @Fayenatic.  This is all early days.  I created the template and set about deploying it on a few series to see how it would work out in practice. My judgement was that it was better than expected, and that the resulting large navbox on large series seemed to work well.  That's why I then posted about it at Template talk:Navseasoncats, in the hope of attracting more heads.
 * I thought from the outset that it might be better used as a meta-template, as you suggest, but I reserved judgement until there were more use cases.
 * However, I think that whether to do so depends on the size of the series, and the frequency/likelihood of new additions or renamings. To my mind, the Scottish MSPs series was too short to need a series-specific template, but I would in no way object to anyone who made a different judgement.
 * And for some larger series like the 15 children of Category:Members of the Karnataka Legislative Assembly by term, I definitely agree that a custom template should be created. However, the name change in that series would need some jiggery in the custom template.
 * Or do you recommend always creating a series-specific template?
 * I think we should see if we can agree on some guidance to add to the documentation. -- Brown HairedGirl  (talk) • (contribs) 10:45, 10 May 2021 (UTC)
 * I'm not sure what would be the threshold for creating a template. As things stand, Scotland is in line to have future MSPs categories… unless it becomes independent, beyond which there could be such a discontinuity that we might not want a nav template to bridge both periods. In that case only one or two future changes would be required for the MSP category series, so the benefit of a template would be marginal. But aside from the possibility of Scottish independence, I would set up a template now.
 * I don't see a naming problem with a Karnataka template. You've set up redirects for Mysore terms already, so the template only needs to use the current name.
 * What might be useful, though, would be an extension of this template to cover a change of prefix (using a named parameter ?), e.g. Welsh AMs to MSs. For that matter, it could handle Mysore to Karnataka without redirects. I guess it would be worth building a sub-template for a line of the current template, to save repeating all the ifexist tests longhand. (That would also simplify extending it later to also handle a change of suffix if required…) – Fayenatic  L ondon 21:54, 12 May 2021 (UTC)
 * Update for the record: Tom.Reding just made navseasoncats work on most cases where this alternative template had been implemented.
 * As it is now only used on Mysore/Karnataka up to Category:Karnataka MLAs 1978–1983, I suggest stopping those nav template lists at 1999. Then they won't need maintaining in future.
 * Then, this will neither be used on, nor include links to, any current-period categories. So, on grounds of future maintenance, there's no current need for any series-specific templates. – Fayenatic  L ondon 18:55, 14 May 2021 (UTC)
 * There's still the problem of variable-term, variable-gap categories like
 * Unfortunately, there is no easy work-around (that I can think of) for these 1-year gaps with no base change (oddly enough, a base change makes it easier), since 2 R from category navigation would be needed in each direction, and I can't think of a logical numbering system that wouldn't circularly conflict with itself in the opposite direction.
 * Fortunately, the most common gap I've seen in is 1 year, and I think I can write that into the existing code. Expensive parse function count is becoming a problem, though, at least in testcase space, where I've so far needed to make up to /testcases3. The highest expensive count I've seen so far after the latest changes (searching for 0-year-gap, variable-term cats), was I think ~52/500. Searching for 1-year gaps would potentially double that to ~100/500 (but more realistically ~75), and 2-year gaps ~125/500. Anything more than a 1-year gap search would make testing prohibitive. (I'm not sure if there's a way to track the expensive function count, but if there is, it would help in assessing the limits of how aggressively Navseasoncats searches.)
 * So Irregular navseasoncats I think is currently useful for these 0-1 year variable-gap, variable-term cats (until (hopefully) superseded by Navseasoncats), and will continue to be useful for 2+ year variable-gap, variable-term cats.  ~ Tom.Reding (talk ⋅dgaf)  20:13, 14 May 2021 (UTC)
 * Update: Navseasoncats should now work on variable terms with variable gaps up to 5 years, and allow for 0-length terms mixed in (e.g. ).  ~ Tom.Reding (talk ⋅dgaf)  19:10, 28 August 2021 (UTC)
 * So Irregular navseasoncats I think is currently useful for these 0-1 year variable-gap, variable-term cats (until (hopefully) superseded by Navseasoncats), and will continue to be useful for 2+ year variable-gap, variable-term cats.  ~ Tom.Reding (talk ⋅dgaf)  20:13, 14 May 2021 (UTC)
 * Update: Navseasoncats should now work on variable terms with variable gaps up to 5 years, and allow for 0-length terms mixed in (e.g. ).  ~ Tom.Reding (talk ⋅dgaf)  19:10, 28 August 2021 (UTC)