Ah looking again, I was mistaken -- lib/test/docs breakdown does not require 3 separate repos. I'll need to take a closer look at how that's set up. Still, the ability to be explicit about package sources could be useful, and may provide a simpler alternative.
On Wed, May 5, 2021 at 8:06 PM Siddhartha Kasivajhula <[email protected]> wrote: > Thank you, that helped me understand the issue. I've made the appropriate > changes so hopefully that'll do it. > > This is reminding me that in python package distribution land, you are > able to explicitly indicate which files make it into the package in a file > analogous to info.rkt (setup.py, but also MANIFEST.in which allows you to > bundle non-source files, I believe). Files besides these simply reside in > the repository but don't make it into either source distributions or > binaries. This could be a desirable feature to emulate in raco, since it > would mean that (1) test modules need not be included as part of a > collection even if present in the repo, (2) I've noticed that some Racket > packages use a lib/test/docs breakdown with 3 separate packages; this seems > non-ideal to me since they need to reside in 3 separate repos even though > the non-source repos are essentially metadata (despite being written in > full-fledged languages like scribble). By supporting something like a > "source-include-paths" info.rkt declaration with a list of paths (similar > to compile-omit-paths and test-include-paths), the tests and docs could all > reside in the same source repo, without their (necessarily) becoming part > of the installed package/collection. Any thoughts on this? > > > > On Wed, May 5, 2021 at 3:35 PM Robby Findler <[email protected]> > wrote: > >> I think the issue is that you should not use "tests/private" as your >> collection path. You should probably either "tests/my-collection/private" >> or "my-collection/tests/private" as the name. >> >> Robby >> >> >> On Wed, May 5, 2021 at 4:38 PM Siddhartha Kasivajhula <[email protected]> >> wrote: >> >>> That makes sense. Is there a recommended way to exclude test paths from >>> being part of the package modules? I have a `compile-omit-paths` >>> declaration in the info.rkt >>> <https://github.com/countvajhula/on-macro/blob/0d861169cbe5b86d70c217be3563809fcae4f3cc/syntax/on/info.rkt#L2-L3> >>> file. Since this is structured as a multi-collection package, could it be >>> that these declarations need to be at the package-level info.rkt file >>> instead of the collection-specific one to take effect on the package index? >>> Another option I can think of is to move the `tests` folder which is >>> currently at the package level down to the collection level, might that >>> work? >>> >>> FTR re: the lazytree package, I assume it's showing conflicts because >>> although it hasn't been changed recently, it depends on the relation >>> package which depends on both on-macro and social-contract, which are in >>> conflict per the above. >>> >>> Re: mischief, I've encountered some weird errors in the past with >>> building docs, for instance see this comment >>> <https://github.com/countvajhula/relation/blob/eba916a37511427f54b9d6093f7620600c99c1a7/function.rkt#L126-L132>. >>> I wonder if that's related to the conflicting names. >>> >>> >>> On Wed, May 5, 2021 at 1:38 PM Sam Tobin-Hochstadt <[email protected]> >>> wrote: >>> >>>> I think there's two things you're seeing. >>>> >>>> 1. The results hadn't yet updated for your typed-compose change. I no >>>> longer see a conflict here: https://pkg-build.racket-lang.org/ >>>> 2. The conflicts page is for _all_ the packages in the whole package >>>> catalog. That's why it always mentions mischief. >>>> >>>> The issue for on-macro and social-contract is that they both have a >>>> file tests/private/util.rkt, which means they can't be installed at >>>> the same time. >>>> >>>> Finally, mischief has that issue intentionally -- there are two >>>> versions of it on the pkg server, one of which is the development >>>> branch. It's true that it hasn't been updated recently, but that's the >>>> idea. >>>> >>>> Sam >>>> >>>> >>>> On Wed, May 5, 2021 at 4:08 PM unlimitedscolobb >>>> <[email protected]> wrote: >>>> > >>>> > Hi, >>>> > >>>> > I'd like to chime back in and say that renaming manual.rkt to >>>> typed-compose.rkt didn't seem to affect much the list of install conflicts >>>> for typed-compose. I also get a lot of conflicts with mischief (but not >>>> only), even though typed-compose doesn't depend on it, or doesn't even >>>> define names which would be similar to what mischief defines. >>>> > >>>> > That's puzzling. >>>> > >>>> > - >>>> > Sergiu >>>> > >>>> > On Wednesday, May 5, 2021 at 9:16:14 PM UTC+2 Siddhartha Kasivajhula >>>> wrote: >>>> >> >>>> >> Hi, >>>> >> I'd like to report that I'm seeing conflicts being reported on my >>>> packages as well. I haven't made recent changes to these packages so the >>>> conflicts seem to have appeared spontaneously. >>>> >> >>>> >> Here is one example: https://pkgs.racket-lang.org/package/lazytree >>>> >> Clicking into the "conflicts" results in a 404. >>>> >> >>>> >> Another example: https://pkgs.racket-lang.org/package/on-macro >>>> >> Here, clicking into "conflicts" seems to implicate, believe it or >>>> not, the `mischief` package, of which it appears there are two separate >>>> versions on the package index. This does seem rather mischievous, and maybe >>>> raco doesn't like it? Yet, it doesn't look like either mischief or >>>> mischief-dev have been changed in years, so I'm not sure why it should >>>> complain now about these longstanding shenanigans. >>>> >> >>>> >> A third example: >>>> https://pkgs.racket-lang.org/package/social-contract >>>> >> Clicking into "conflicts" once again seems to implicate mischief, >>>> but mischief isn't even in the dependencies for this package so this just >>>> seems unfair! >>>> >> >>>> >> On other packages that I've uploaded, the conflicts link was a 404. >>>> >> >>>> >> Similar questions as OP - should I fix something here, for instance >>>> by avoiding the mischief dependency? Should mischief itself be updated in >>>> some way? Or is this (as seems likely) caused by a recent change in the >>>> package index, and if so, how should package authors respond (assuming it >>>> isn't a bug)? What to do about the 404s -- e.g. is there a command to >>>> generate the conflicts locally? >>>> >> >>>> >> Thanks, >>>> >> >>>> >> >>>> >> >>>> >> On Sun, May 2, 2021 at 6:59 AM unlimitedscolobb < >>>> [email protected]> wrote: >>>> >>> >>>> >>> Hi Jay, >>>> >>> >>>> >>> Thanks a lot for helping me read that file! >>>> >>> >>>> >>> I didn't know Scribble outputs shared the same namespace. I >>>> renamed the documentation file to typed-compose.scrbl as you suggest and >>>> I'm waiting for build reports from the package catalog. >>>> >>> >>>> >>> In fact, I hesitated between manual.scrbl and typed-compose.scrbl >>>> initially, and couldn't find a reason to prefer one over the other. Now I >>>> have a reason :-) >>>> >>> >>>> >>> - >>>> >>> Sergiu >>>> >>> >>>> >>> On Saturday, May 1, 2021 at 3:23:47 PM UTC+2 [email protected] >>>> wrote: >>>> >>>> >>>> >>>> Howdy Sergiu, >>>> >>>> >>>> >>>> The conflicts file you link to has all the conflicts for everything >>>> >>>> that `pkg-build` builds. The line relevant for you is: >>>> >>>> >>>> >>>> ``` >>>> >>>> doc "manual": >>>> >>>> bystroTeX cbor print-debug ratchet riff simply-scheme typed-compose >>>> >>>> ``` >>>> >>>> >>>> >>>> The solution is to rename your manual from `manual.scrbl` to >>>> >>>> `typed-compose.scrbl`. Scribble outputs are in a kind of "global" >>>> >>>> namespace. >>>> >>>> >>>> >>>> Jay >>>> >>>> >>>> >>>> -- >>>> >>>> Jay McCarthy >>>> >>>> Associate Professor @ CS @ UMass Lowell >>>> >>>> http://jeapostrophe.github.io >>>> >>>> Vincit qui se vincit. >>>> >>>> >>>> >>>> >>>> >>>> -- >>>> >>>> Jay McCarthy >>>> >>>> Associate Professor @ CS @ UMass Lowell >>>> >>>> http://jeapostrophe.github.io >>>> >>>> Vincit qui se vincit. >>>> >>>> >>>> >>>> >>>> >>>> On Sat, May 1, 2021 at 5:56 AM unlimitedscolobb >>>> >>>> <[email protected]> wrote: >>>> >>>> > >>>> >>>> > Hello, >>>> >>>> > >>>> >>>> > I checked my package >>>> https://pkgd.racket-lang.org/pkgn/package/typed-compose recently and >>>> noticed that it listed some "Conflicts" in the field "Most recent build >>>> results". On the other hand, the separate field "Conflicts" slightly above >>>> says "None". >>>> >>>> > >>>> >>>> > When I open the log shown in "Most recent build results" >>>> (attached) it starts with the line "Install conflicts:", which as far as I >>>> get are not the same thing as "Package Conflicts" explained here in the >>>> manual: >>>> https://docs.racket-lang.org/pkg/Package_Concepts.html#(part._concept~3aconflicts) >>>> . >>>> >>>> > >>>> >>>> > What are install conflicts? Should I fix them? What is the >>>> command that generates that log? >>>> >>>> > >>>> >>>> > (typed-compose also used to have an undeclared dependency, which >>>> I have just fixed by updating it's info.rkt.) >>>> >>>> > >>>> >>>> > - >>>> >>>> > Sergiu >>>> >>>> > >>>> >>>> > -- >>>> >>>> > You received this message because you are subscribed to the >>>> Google Groups "Racket Users" group. >>>> >>>> > To unsubscribe from this group and stop receiving emails from >>>> it, send an email to [email protected]. >>>> >>>> > To view this discussion on the web visit >>>> https://groups.google.com/d/msgid/racket-users/03bfa876-6418-41a2-a37a-5c39ad13121cn%40googlegroups.com >>>> . >>>> >>> >>>> >>> -- >>>> >>> You received this message because you are subscribed to the Google >>>> Groups "Racket Users" group. >>>> >>> To unsubscribe from this group and stop receiving emails from it, >>>> send an email to [email protected]. >>>> >>> >>>> >>> To view this discussion on the web visit >>>> https://groups.google.com/d/msgid/racket-users/eb380f03-7a73-40e7-a3a6-0e8c3252525an%40googlegroups.com >>>> . >>>> > >>>> > -- >>>> > You received this message because you are subscribed to the Google >>>> Groups "Racket Users" group. >>>> > To unsubscribe from this group and stop receiving emails from it, >>>> send an email to [email protected]. >>>> > To view this discussion on the web visit >>>> https://groups.google.com/d/msgid/racket-users/310449db-1450-4043-9d73-c80c95587653n%40googlegroups.com >>>> . >>>> >>>> -- >>>> You received this message because you are subscribed to the Google >>>> Groups "Racket Users" group. >>>> To unsubscribe from this group and stop receiving emails from it, send >>>> an email to [email protected]. >>>> To view this discussion on the web visit >>>> https://groups.google.com/d/msgid/racket-users/CAK%3DHD%2BZVw3eRSgvw_C5K%3DDY0Ndiy1v5pnv5j8DeUgaeCvJYCHg%40mail.gmail.com >>>> . >>>> >>> -- >>> You received this message because you are subscribed to the Google >>> Groups "Racket Users" group. >>> To unsubscribe from this group and stop receiving emails from it, send >>> an email to [email protected]. >>> To view this discussion on the web visit >>> https://groups.google.com/d/msgid/racket-users/CACQBWFmJ5eKH0cHcnKzGge0h1wcuMoz_MUFRP%3D%3DuZ8_%3D98Pikw%40mail.gmail.com >>> <https://groups.google.com/d/msgid/racket-users/CACQBWFmJ5eKH0cHcnKzGge0h1wcuMoz_MUFRP%3D%3DuZ8_%3D98Pikw%40mail.gmail.com?utm_medium=email&utm_source=footer> >>> . >>> >> -- You received this message because you are subscribed to the Google Groups "Racket Users" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. To view this discussion on the web visit https://groups.google.com/d/msgid/racket-users/CACQBWFn6gbxZJbPxLD0im9J8cDQ7W3uKyk2R%3D%2BAGanwBeii%2B1Q%40mail.gmail.com.
