Add `tag_level = "stop"`

[thomasp85]

Stop descending deeper into the nesting and only add a single tag to the whole nested patch

[Yunuuuu]

Why not let each patchwork control its own tag independently, instead of having the top-level patchwork object control all tag_levels? That would make the system more flexible.

[thomasp85]

Because you often want tagging to descend into nested layouts

[Yunuuuu]

I mean something like this: https://yunuuuu.github.io/ggalign-gallery/basics/plot-composer-tagging.html

If not defined in the nested layout, we just use the parent layout tags.

In this way, we can define tag prefix and suffix for each patchwork even for nested layouts.

align_plots(
    p1,
    align_plots(p2, p3) + layout_tags(1, sep = ": "),
    ncol = 1
) +
    layout_tags("A", prefix = "Fig.")

[thomasp85]

I'm afraid I don't see the benefit of this. It would for the most part lead to more typing and the need to change things multiple places if you wish to have a different enumeration

[stitam]

While I understand the value of automating tag generation, I keep bumping into edge cases. For example, in my current project I need to have a single tag for two figures (technically a patchwork figure, but semantically they are treated as a single figure in the text), and then a second tag for a (technically) third figure.

I don't want to focus on the issue itself, I'm just saying there will always be edge cases. While I currently use patchwork for composing figures, in cowplot we can just define tag names as a character vector which I find very useful. At this point tag names are the only reason why I sometimes need manual rework on my patchwork composite figures.

@thomasp85 have you considered implementing manually defined tags? I don't think it should be super difficult, I think the rule "if the length of the vector is the same as the number of figures, treat them as manually defined tags" should work because in automated situations I think you'll always have fewer strings.

[stitam]

Short update, I found that manual tags are possible (though maybe not super intuitive):

p1 + p2 + p3 + plot_annotation(tag_levels = list(c("tag1", "tag2", "tag3")))

Also if I want to leave out the second tag:

p1 + p2 + p3 + plot_annotation(tag_levels = list(c("tag1", "", "tag3")))

We have to wrap the character vector into a list.