- Only print "up-to-now" theorems, or by section.
This is complicated. See etoc. And acro (
\acbarrier
). - (just use
\theoremstyle
) Use style key for multiple theorems, liketex \keytheoremset{style={<keys>}} \newkeytheorem{thm1} \newkeytheorem{thm2}
Well maybe not. Because should these style keys also be applied to a called\newkeytheoremstyle
? Or only used for that? Question is: should user be able to style theorems without an explicit call to\newkeytheorem{<thm>}[style=<style>]
? - link theorem to restated (see TeX.sx)
- Fix equation, etc. numbering in restated theorems when numbered by chapter, section, etc. Or leave it up to user to add these counters with restate-counters?
- listhack
- Proof hooks? Other proof customization? New proof-like envs? feature creep...
-
\newtheoremstyle
overwrites existing styles. Should\newkeytheoremstyle
check if style exists? Relevant for plain, remark, definition. Could provide\renewkeytheoremstyle
. Further, if there is use for (re)defining theorems mid-document, could make cmds usable after preamble by removing package loading in keys Slightly more complicated than anticipated, need to deal with hooks - Currently
numbered=unless-unique
+parent
is incompatible with restate, but this is true also with thmtools. - Idea for qed, tcolorbox in style: in thm, just check if already set and adjust accordingly
- What about
\zlabel
and other "label" commands in restated theorem? Should there be an interface for disabling them? -
unless-unique
more general: https://tex.stackexchange.com/a/705572/208544 - Rename "preheadhook" etc. to "prehead"? Answer: no, prehead sounds like a skip amount
-
\listofkeytheorems
does not print restated theorems. Should an option be added to do this? No, right? - Language support? At least for list of theorems title
- Interface for adding entries to list of theorems. In fact, should
benchmark the "add to seq then write at end of file" approach vs.
traditional
\addcontentsline
approach -
short name
key for theorems to replace thmtools'name={[short name]name}
. Should we also support this syntax? No. - (see restate-keys) And what about
restate={[options]foo}
syntax? How useful is this? - Should
continues(*)
theorems appear in list of theorems? Decide and add option to change default. - Add
shaded
andthmbox
keys that use tcolorbox under the hood. In thmtools-compat or available always? Or as "library"? Currently thmbox has white background, want transparent but has issue: https://tex.stackexchange.com/a/706216/208544 - Should more hook operations be available in user commands, like labels and removing?
Answer: no, I don't think so.
\addtotheoremhook
is for simple things. Programmers can use the usual hook interface with more verbose labels - Document how to disable things like footnotes in restated theorems.
- Should indent be suppressed after tcolorbox theorems? It is for thmtools' shaded and thmbox theorems.
Just setting
\tcbset{after=\par\@endpetrue}
doesn't work since there is code in between but\AddToHook{keytheorems/<env>/postfoot}[keythms_hook_keys]{\par\@endpetrue}
seems to work Answer: I say no, since it's not suppressed after regular theorems. - (partial)
break
+tcolorbox-no-titlebar
adds too much space if theorem starts with list listhack fixes this. - rename package to "keytheorems"?
- should all code added to hooks have the "keytheorems" label as opposed to "."?
- seq approach doesn't work well with
\includeonly
- chaptervspacehack
- Should there be a way to add content lines to
\listofkeytheorems
with print-body? Currently we disable code added with\addtotheoremcontents
- hyperref does not jump to correct location for tcolorbox theorems
- use
\MakeLinkTarget
instead of dummy counter? - should default theorem style be changed if ams class is loaded? Yes, and need general mechanism for setting up class defaults
- need to clear .thlist file if no restate or list of
- Beamer issues:
- tcolorbox theorems not compatible with beamer
- cannot give action spec to theorems
- Idea: delay making hooks until begindocument to avoid using internal cmd
- fix acmart defaults
- use different dummy counters for unnumbered, continues, and restate
- seq for custom lists of theorems
- make code more modular
- option to restate without writing to file (let's wait to see if this is an issue)
- code with brackets fail in note key, e.g.
\cite[bla]{ref}
- with tcolorbox theorems, spaceabove and spacebelow should always be set to 0pt
- no-auto-translate or something like that
- tcolorbox theorem spacing after sections is not good when titlesec loaded
- also different behavior for tcb theorems and not, namely spaceabove adds space after section for former but not latter
- (see
format-code
) full control over list-of appearance, including name and number - option to add list of theorems to TOC. No, let user control manually.
- list appearance does not match list of figures in memoir
- make indent configurable in general
- With amsart etc.
no-title
vert spacing is bad; currently uses empty title - chaptervspacehack adds extra vert space above no-title (maybe this is okay)
- With memoir, list of theorems always on new page (list of figures is not) Should I just use memoir's list commands when it's loaded?
- Idea: just define everything for the default classes, then have support files for the other classes that redefine things
- Remove dependency on unique.sty
- Line numbers for errors/warnings
- store-sets-label
- allow passing options to restated env
- number font
- allow to suppress content of env, then restate later (or earlier)
-
\begin{theorem}[store=foo][text that should be in body]
picks up latter bracketed material as note -
\theoremstyle
does not apply thm keys - nested inherit-style errors
- use tcb style with #1 for tcolorbox theorems so style can be added to or overridden
- can't define
\renewkeytheorem
, etc., until figure out preheadhook situation. Right now we don't add if empty, but then \hook_remove gives warnings -
\KeyThmsContentsLine
, etc. should be default be no-ops, redefined to do things, not reverse as now - amsbook and
no-title
... - hook order with
\addtotheoremhook
and qed is wrong (or is it?) -
\getkeytheorem[body]{foo}
needs to use theorem-specific restated hook This means getthm_body needs to take three arguments... -
qed={}
in style does not do the right thing - lots of code duplication, for example in beamer support file
- space after title with beamer, tcolorbox theorems, and label
- move
prehead_code
out ofwithhooks_begin
so the preheadhook can change note. Then wouldn't need separateprehead_continues_code
, I think. See this for application: https://tex.stackexchange.com/a/728812/208544 - logic with
restate-keys
andstore*
is reversed. Nevermind, it has to work this way. Otherwise all keys for actually written theorem would have to be passes torestate-keys
. - Don't patch
\@thm
forleftmargin
andrightmargin
. Just redefine\trivlist
in the prehead hook and restore original definition in posthead. - Make margin theorems work with tagging code
- Manual label after
tcolorbox-no-titlebar
theorem produces extra space - Nested theorem inside tcolorbox theorem
- Default spacing of tcolorbox theorems
- For tcb-no-title, wrap in prehead and postfoot hooks and just declare hook rules to match tcolorbox key. Otherwise this is a pain
- Decrease garishness of documentation
- for thm keys used in
\newkeytheoremstyle
,\l_keys_key_str
is wrong wheninherit-style
is called - adjust doc for spaceabove, etc. that are skips, not dimens
-
continues*
does not work with manual\label
and nameref not loaded - theorem option to manually override label
- full example in doc, including restating theorem in intro before stated
- allow customization of listhack skip
-
onlynamed
andcontinues