Closed miguel-negrao closed 11 years ago
I think I got it:
While in "install" the quark version is not checked, in fetchQuark it is checked with pr_chooseBestVersion which returns Dictionary[ (compat -> [ 3.4, 3.5, 3.6, 3.7 ]), (fetchInfo -> v0.1), (version -> 0.1) ] . So if giving no "quarkversion" means "give me the latest" it should still return nil, instead of v0.1, should it not ?
I think it would be good if the latest version can also be entered manually in the yaml file so that the quark maintainer can specify the compatability of the latest version. If the latest version is only compatible with 3.6, and you are in on 3.5 you don't want to install the latest vesion, yet if giving no version info is supposed to mean that we want the latest it will cause the instalattion of an incompatible version. Perhaps we can allow for a version "latest" to be manually entered in the yaml file ? If it is not entered we assume latest is compatible with every sc version.
Ah yes - the logic was flawed, thanks for pointing it out. Commit 46a31b8f87e7a9a2 should fix it. (If you can confirm, I'll close the issue.)
The quark maintainer can already specify the compatibility of the latest version. It simply goes in the root.
Ok, all is working good now ! Witness the firepower of this fully operational Quarks2 implementation ! ;-) WIll convert my yaml files version -> versions.
Doing Quarks2.install("Modality-toolkit") is checking out v0.1 in git instead of HEAD. It is naming the folder as Modality-toolkit-3.7-latest though...