Build features #41
Open
Build features #41
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This anticipates the statx() interface I'll add soon. Breaking change: Adds a 'SimpleType::Unknown' which can be returned when the filetype can't be determinded. Code matching on SimpleType may add a clause for this when there is no catch-all.
dir.list() consumes the dir and creates a DirIter. The underlying 'dup()' operation now becomes explicit, one may need to call try_clone() first. the list_self() and list_dir() using this now but are merely convinience wrappers and may (or may not) deprecated in future. Rationale: this simplifies the code a bit (list::open_dir() removed) and is closer to the low level semantics.
* Forgot to forget the Dir handle, thus it would been dropped. * On linux O_PATH was passed when Dir descriptors where opened. This is generally a good thing but broke the refactored list(). This also shown that O_PATH has different enough semantics to be problematic vs. opening handles without (as on other OS'es). Changed this to Dir::open() and Dir::sub_dir() default to opening without O_PATH (but O_DIRECTORY see remarks). * added Dir::open_path() and Dir::sub_dir_path() with O_PATH (urgh, better idea for naming? open_protected() or something like that?) which offer the O_PATH feature on linux.
'Lite' file descriptors are possibly a better naming of what O_PATH does. This introduces then and implements them portable. On systems which do not support O_PATH normal file descriptors are used.
* With FdType/fd_type() one can determine the kind of an underlying file descriptor. Lite descriptors are implemented only in Linux (for now). When O_DIRECTORY is supported it uses fcntl() in favo over stat() * clone_dirfd() tries to do 'the right thing' for duplicating FD's * libc_ok() is a simple wraper for libc calls that return -1 on error, I will refactor the code in the next commits to make use of that more. Please test this! So far it works for me on Linux.
This simplifies the code a bit, in 'release' the same output is generated. debug builds may not inline the libc_ok() and be slightly larger.
This up/downgrade cloning converts into normal/lite handles which was missing before. I hope this fixes tailhook#34 finally.
the d_ino field is mandatory in POSIX and clients (me) really need it when passing data around. Keeping it in the Entry saves an extra stat()/metadata query.
Instead compiling conditionally on the target_os this defines features for every aspect that can be different. Tested only on linux so far. Eventually the presence/absence of these features should be determined by a build.rs script. Meanwhile there are OS specific lists (only linux so far, please contribute more) which can be activated when building with '--feature osname'.
|
Works now. this adds some more build dependencies to run a build.rs script now and a intermezzo on myself created https://docs.rs/conf_test/latest//conf_test/index.html for configuration checks (missed that functionality anyway rust/cargo). Initial build time got a bit up now, but that's only a a once only overhead. OS Features and capabilities are now checked from the code snippets in conf_tests/*.rs and enabled automatically. Would be nice when people could test this on other (non linux/non macos) operating systems because it may happen that the tests are not 'on the spot' yet. For example that older OSX supports O_DIRECTORY but fcntl() won't return that flag left me initially puzzled. Newer OSX does not seem to have this behavior. Anyway when problems are known its pretty simple now to to factor a 'feature' of of that and add a test enabling that feature. |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
This is a draft so far, it builds on top of my other PR's. Eventually this should make the library configuration much simpler. I would appreciate any help in this area since I am primary only interested in Linux can cant test elsewhere.