Zulip Chat Archive

Stream: mathlib4

Topic: Moving Filter


Yaël Dillies (Jan 17 2024 at 19:17):

Mind moving all the filter under the Topology folder btw?

Yury G. Kudryashov (Jan 17 2024 at 19:28):

Yaël Dillies said:

Mind moving all the filter under the Topology folder btw?

I don't want to move files around without prior feedback from @Mario Carneiro on how it affects Mathport and his future leanup tool.

Yury G. Kudryashov (Jan 17 2024 at 19:29):

We had Filter under Order for a long time, we can wait a few more months until we have some tools/decisions about backward compatibility and renames.

Mario Carneiro (Jan 18 2024 at 02:47):

Yury G. Kudryashov said:

Yaël Dillies said:

Mind moving all the filter under the Topology folder btw?

I don't want to move files around without prior feedback from Mario Carneiro on how it affects Mathport and his future leanup tool.

The usual advice is still in effect here: if you want to split a file, clone the #align_import line into all the fragments, and if you want to join a file put two #align_import lines in the file. As for leanup, I wouldn't worry too much about it until we have a working system, because all of this will be considered "historical data" from leanup's perspective and we'll do the best we can with whatever is available. (leanup will probably make use of #align and #align_import to track renames and splits in historical data, because it's been around for a while and is relatively accurate.)

Yury G. Kudryashov (Jan 18 2024 at 03:15):

It's about moving Order/Filter to Topology/Filter, not splitting (I know what to do with splitting).

Kim Morrison (Jan 18 2024 at 03:22):

I think you can just leave the #align's as is, then, and everything will be fine re: mathport.

Yury G. Kudryashov (Jan 18 2024 at 04:04):

@Yaël Dillies If you want this to happen, then please open a new topic. I don't care where docs#Filter is located but other maintainers can have reasons not to move it.

Mario Carneiro (Jan 18 2024 at 04:05):

It's not obviously a better place for it to me

Mario Carneiro (Jan 18 2024 at 04:07):

a filter is kind of like a topology but they are sister concepts more than parent/child. PFilter is even more order theoretic

Yury G. Kudryashov (Jan 18 2024 at 04:08):

Let's postpone this discussion (offtopic for this thread) until Yael opens a new topic.

Yaël Dillies (Jan 18 2024 at 08:22):

Sorry it's annoying to move messages from phone (/impossible?) so I'll reply here: I want Filter to become Topology.Filter and Order.PFillter to become Order.Filter. Eventually, Topology.Filter will just be notation around Order.Filter and will be deletable or replaceable by notation.

Mario Carneiro (Jan 18 2024 at 08:41):

that seems... confusing

Yaël Dillies (Jan 18 2024 at 09:14):

Why so?

Notification Bot (Jan 18 2024 at 09:15):

A message was moved here from #mathlib4 > Linter for large files? by Yaël Dillies.

Mario Carneiro (Jan 18 2024 at 09:15):

you are giving two different definitions the same name

Yaël Dillies (Jan 18 2024 at 09:17):

That's why there are namespaces involved. Note that they have the same name in the literature already.

Mario Carneiro (Jan 18 2024 at 09:24):

Still annoying, since either you have a really long name for one of them or you open both and have name conflicts

Mario Carneiro (Jan 18 2024 at 09:24):

and more problematically, you have to look at an open halfway up the file when reading code to know which is being talked about

Mario Carneiro (Jan 18 2024 at 09:26):

Generally, I think the unqualified names of things in the same or adjacent domains should be distinct

Mario Carneiro (Jan 18 2024 at 09:26):

namespacing things with the same unqualified name is most useful when the same word is used for unrelated things in unrelated domains


Last updated: May 02 2025 at 03:31 UTC