-
Notifications
You must be signed in to change notification settings - Fork 697
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
toConfiguredComponent: unix (Windows) #3978
Comments
@ezyang, can you please look into this? |
@alexbiehl Any chance I can get some -v logs? (Or better yet, -v3) |
Yes of course, I will hand them in tomorrow. Edward Z. Yang [email protected] schrieb am Do., 13. Okt. 2016,
|
And here the same command as a subsequent run (with a I have one question to add: Somewhere in the logs it says |
new-build will solve for ALL local packages in your project, so that when you decide to build some of the other ones, it doesn't have to resolve. |
If you can, apply this patch and then give me the non-verbose log?
|
|
So, I guess this is not a direct dependency of hunt-searchengine, but one of the other packages that are local? |
Yes, right. |
I guess I have some guesses how I might reproduce this locally. I guess this is not urgent, since you can just remove the unbuildable packages from your project? |
No not urgent. I already filled a pull-request for |
OK, the bug seems to be somehow related to
Stick this in a package with a building package (say |
OK, so here's the problem: when the library of a pckage is marked as non-buildable, new-build does NOT eagerly fail. In 1.24, the error would only happen if you tried to build it, or anything that depended on it:
or
I suppose that's reasonable behavior; it lets you build some part of your project even when other parts are non-buildable. But when I refactored this code for Backpack, I assumed that I would only ever get plans with packages that depended upon buildable things. Thus the assert failure! The library doesn't exist and never was configured. I guess one way we could fix this is to have the solver reject a package if the library is marked as non-buildable (as opposed to executables/tests/etc. May god have mercy on your soul if you have a non-buildable library with a buildable executable.) But maybe this is not so good for you because it means that you can't develop a project if any of the local deps transitively depends on a non-buildable thing. But maybe you didn't care so much about this case? By the way, |
Now you get something like this: Resolving dependencies... cabal: Could not resolve dependencies: trying: p-0.1 (user goal) unknown package: unbuildable (dependency of p-0.1) fail (backjumping, conflict set: p, unbuildable) Dependency tree exhaustively searched. The error message is not the best. Fixes haskell#3978 Signed-off-by: Edward Z. Yang <[email protected]>
@ezyang fyi, I've seen the same error |
@dcoutts Do you have a repro? |
When a library/executable is unbuildable, the solver may still return a reference to it. Handle this gracefully in ProjectPlanning with a nice error message, rather than an assert failure. Fixes haskell#3978. Signed-off-by: Edward Z. Yang <[email protected]>
When a library/executable is unbuildable, the solver may still return a reference to it. Handle this gracefully in ProjectPlanning with a nice error message, rather than an assert failure. Fixes haskell#3978. Signed-off-by: Edward Z. Yang <[email protected]>
When a library/executable is unbuildable, the solver may still return a reference to it. Handle this gracefully in ProjectPlanning with a nice error message, rather than an assert failure. Fixes haskell#3978. Signed-off-by: Edward Z. Yang <[email protected]>
When a library/executable is unbuildable, the solver may still return a reference to it. Handle this gracefully in ProjectPlanning with a nice error message, rather than an assert failure. Fixes haskell#3978. Signed-off-by: Edward Z. Yang <[email protected]>
I get this error when
with |
Now I see the problem: http://hdiff.luite.com/cgit/http-client/diff?id=0.5.12.1&id2=0.5.12 clicksaver
|
I opened an issue for the new type of failure: #5325 |
Now you get something like this: Resolving dependencies... cabal: Could not resolve dependencies: trying: p-0.1 (user goal) unknown package: unbuildable (dependency of p-0.1) fail (backjumping, conflict set: p, unbuildable) Dependency tree exhaustively searched. The error message is not the best. Fixes haskell#3978
Trying
new-build
with cabal-install from HEAD on a project of mine:Results in:
I would
cabal
expect to fail on buildingunix
(since I am on Windows) instead of some internal error.The text was updated successfully, but these errors were encountered: