summaryrefslogtreecommitdiff
AgeCommit message (Collapse)Author
2020-07-04more quotingJoey Hess
2020-07-04propellor spinJoey Hess
2020-07-04better approach for finding hub port that works when port is powered offJoey Hess
2020-07-04Revert "be robust against hub location id changes"Joey Hess
This reverts commit 70e503e724c3c8cb4312cd128535afc1e06c84a8. Bad approach because the device vendor and id are not visible when the port is turned off.
2020-07-04propellor spinJoey Hess
2020-07-04one moreJoey Hess
2020-07-04propellor spinJoey Hess
2020-07-04more fixquotingJoey Hess
2020-07-04propellor spinJoey Hess
2020-07-04fix shell quotingJoey Hess
2020-07-04propellor spinJoey Hess
2020-07-04be robust against hub location id changesJoey Hess
2020-06-21comment typoJoey Hess
2020-06-18propellor spinJoey Hess
2020-06-18propellor spinJoey Hess
2020-06-18propellor spinJoey Hess
2020-06-18propellor spinJoey Hess
2020-06-18propellor spinJoey Hess
2020-06-18todoJoey Hess
2020-06-18propellor spinJoey Hess
2020-06-18setting up joeyconfig after mergeJoey Hess
2020-06-18simplify update of propellor binJoey Hess
propellor won't be a symlink to propellor.built, instead the latter is renamed over the former, atomically. If something somehow depends on propellor being a symlink, this will break it, but I don't think anything should.
2020-06-18fix one more hardcoded dist/ pathJoey Hess
Refactored the cabal exec code to reuse it here. Had to change the location of propellor.built too, so put it in the top of the propellor repo, rather than in dist. This seems like it could be simpler. Why does propellor need to symlink to propellor.built? Why not copy to a temp file, and atomically rename it to propellor? It seems possible that something could depend on it being a symlink, but I don't know what. Adding to my confusion, there's the comment about cp -pfRL and timestamp checking. What timestamp checking? The commit adding the comment didn't change anything else, and I can't find any now or then. This commit was sponsored by Jack Hill on Patreon.
2020-06-18remove outdated noteJoey Hess
2020-06-18use cabal exec here tooJoey Hess
2020-06-18stop using dist/setup-config as indication propellor is configuredJoey Hess
cabal is changing to new-dist, and does not write setup-config in the same place. Instead, use a top-level "configured" stamp file. Same as the Makefile does now. This will lead to one extra run of cabal configure on each host, in order to get the new stamp file written. This commit was sponsored by Jochen Bartl on Patreon.
2020-06-18use cabal exec to find path of propellor binaryJoey Hess
This turns out to be much faster than cabal install, which does unnecessary rebuilds and other work (https://github.com/haskell/cabal/issues/6919) This commit was sponsored by Jack Hill on Patreon.
2020-06-18Merge branch 'joeyconfig'Joey Hess
2020-06-18tail the dist-newstyle findJoey Hess
Unfortunately builds for previous versions can linger in there. Which may be a cruft accumulation problem generally, but for now I don't want it to fail if more than one is found. Assuming that the last item in the find will be the newest. This commit was sponsored by Brock Spratlen on Patreon.
2020-06-18(no commit message)david
2020-06-17commentJoey Hess
2020-06-17clean up after mergeJoey Hess
2020-06-17Merge branch 'joeyconfig'Joey Hess
2020-06-17Support bootstrapping to hosts using cabal 3.x, with new-dist directory.Joey Hess
* Support bootstrapping to hosts using cabal 3.x, with new-dist directory. * Makefile: Fix build with cabal 3.x. This assumes that, once a new-dist directory is created, the host won't revert back to using dist. So it always prefers binaries from new-dist over dist. This commit was sponsored by LND on Patreon.
2020-06-11Added a comment: network-manager snippetdavid
2020-06-11Added a comment: prototypedavid
2020-06-10Added a comment: mDNS is okdavid
2020-06-10propellor spinJoey Hess
2020-06-10yay, back on track to fix this in um ... 3 yearsJoey Hess
2020-06-10propellor spinJoey Hess
2020-06-10propellor spinJoey Hess
2020-06-10merge from concurrent-outputJoey Hess
This includes e6d4139e15b3a52f4a60178bb7d15ba96f191340 which I hope fixes the reversion that has been plaguing propellor when trying to use more recent versions of concurrent-output. Embedding it will let me test, and also it will be years until that fix is widely enough available to depend on it.
2020-06-10moreJoey Hess
2020-06-10mehJoey Hess
2020-06-10Revert "Added dependency on concurrent-output; removed embedded copy."Joey Hess
This reverts commit dbd3ba3400a3097498252097540ffe8075b00833. Still has the same problem as in 2018!
2020-06-10propellor spinJoey Hess
2020-06-10propellor spinJoey Hess
2020-06-10propellor spinJoey Hess
2020-06-10propellor spinJoey Hess
2020-06-10propellor spinJoey Hess