Zulip Chat Archive
Stream: mathlib4
Topic: naming convention: indexed propositions
Eric Wieser (Feb 16 2024 at 10:02):
For the indexed supremum and union, we have iSup
and iUnion
.
What name do we want for indexed propositions? We currently have:
- docs#LinearMap.IsOrthoᵢ (doesn't match
iSup
, and doesn't linkify on Zulip) - docs#ProbabilityTheory.iIndepFun, docs#ProbabilityTheory.iIndepSet (breaks the
CamelCase
rule forProp
)
Eric Wieser (Feb 16 2024 at 10:11):
docs#Finset.disjiUnion has a relatedly-mangled name, where the i
for "indexed" is lower case and incorrectly reads as part of disj
Yaël Dillies (Feb 16 2024 at 10:12):
I argued before this one should be called iDisjUnion
Eric Wieser (Feb 16 2024 at 10:13):
What's the rationale for choosing that over disjIUnion
?
Yaël Dillies (Feb 16 2024 at 10:14):
That one-letter modifiers should always come first
Eric Wieser (Feb 16 2024 at 10:15):
/poll What should orthogonality of a family of vectors be called
- LinearMap.IsOrthoᵢ
- LinearMap.IIsOrtho
- LinearMap.iIsOrtho
Yaël Dillies (Feb 16 2024 at 10:15):
Similarly, people agreed with me before that docs#Filter.limsSup should be Filter.sLimsup
Eric Wieser (Feb 16 2024 at 10:15):
(same question applies to the indexed IndepFun
and IndepSet
and CondIndep
and ...)
Antoine Chambert-Loir (Feb 16 2024 at 11:16):
Can you explain the rationale for the capitalized initial in the poll ?
Eric Wieser (Feb 16 2024 at 11:18):
#naming says that all types (which includes Prop
s) should be CamelCase
Antoine Chambert-Loir (Feb 16 2024 at 11:23):
What to do with the present inconsistencies regarding #naming, then? For example docs#ProbabilityTheory.iIndepSet
Yaël Dillies (Feb 16 2024 at 11:24):
(Antoine, have you read the start of the thread?)
Antoine Chambert-Loir (Feb 16 2024 at 11:24):
(Yes, this is why I ask.)
Eric Wieser (Feb 16 2024 at 11:27):
The poll above is assuming we change everything to be consistent, and asking which choice we want to have everywhere
Eric Wieser (Feb 16 2024 at 11:27):
Hence this message
Antoine Chambert-Loir (Feb 16 2024 at 11:31):
(I explain my vote: I see the first letter s
, i
, etc. as a modifier, somewhat distinct from the rest of the type. Also, double I is not legible, and I usually prefer to avoid Unicode characters.)
Eric Wieser (Feb 16 2024 at 11:32):
If we go for the i
prefix, then the outcome will also be to change #naming I guess
Antoine Chambert-Loir (Feb 16 2024 at 11:36):
(unless you wish to perpetuate this kind of discussion, yes…)
Eric Rodriguez (Feb 16 2024 at 11:44):
I guess it's nice having it be lower-case at the start - it makes it super easy to search for indexed propositions (when autocomplete finally works...)
Last updated: May 02 2025 at 03:31 UTC