reconsider flush placement #1
Comments
|
fixed in the last commit |
seantcanavan
added a commit
that referenced
this issue
Nov 14, 2016
…ed files. finally figured out this is the best way to store external files to access across any package. having no files local and all files in the asset folder significantly reduced complexity of loading files from any package in any test / execution scenario. 02) updated the makefile reference to version.no to the new /assets location 03) updated the .gitignore file to not look for the new gmail passwords file 04) moved all .json files to ../assets/ to keep the folders organized and tidy 05) removed shell expansion variables from the profiler_loader configuration files as shell expansion does not work when executing commands from go. need to re-think how to do this. 06) removed the reference to LOCAL_EXTERNAL_PATH in config.go. resources are now resolvable by just their raw name using utils.AssetPath() which will perform all the wizardry necessary in order to locate them. 07) cleaned up the config interface to be more consistent with native go code 08) updated the clarity of the errors in config_test.go to better accommodate the asset load and the config load 09) removed all references to 'get' to be closer to native go code 10) removed all extra config files which were added recently and were a lazy workaround for not being able to resolve the same file by relative paths across different packages. see #1 for more. 11) added the ability to load OS-specific variant files with the utils package and SysAssetPath. It inserts the current platform into a filename before the extension to au11) added the ability to load OS-specific variant files with the utils package and SysAssetPath. It inserts the current platform into a filename before the extension to au11) addedt i11) added the ability to load OS-specific variant files with the utils package and SysAssetPath. It inserts the current platform into a filename before the extension to au11) added the ability to load OS-spe less spaghetti now. could be shorter - probably. but possibly at the cost of readability. 14) updated reporter to remove the global instance. if we were to all the variables we need directly from Cfg we don't need the struct anymore. this has the benefit of getting the most up-to-date values for sending out emails after they've been updated via REST (in the future). everything is static now and the interface should be clear. 15) updated most TestMain functions across packages to correctly use asset to load the config file and test if there was an error. before we were not checking on every config load if an error actually returned or not. this was causing nil dereference panics. 16) correctly storing logger now inside of rest. the entire point of this push was supposed to be to finish the rest_test.go but got sidetracked... again... oh well results are great. 17) small work towards unifying references to the same packages with the same 3 letter names: Profiler -> pfr Logger -> Lgr Loader -> Ldr, etc. more work to be done here i'm sure. 18) beginnings of rest_test were created. need to work on timestamp verification and hitting the endpoint in test. for now the server is coming up correctly and generating a random port AND emails are sent out now when the port is up! 19) added unit tests for utils.AssetPath and utils.SysAssetPath
seantcanavan
added a commit
that referenced
this issue
Sep 25, 2017
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
when reading the logs for the profiler test we needed to ensure they were flushed from the writer before being able to peek at the log file to pull from it. i lazily put .flush after every individual log output which defeats the purpose of buffering in the first place. we should probably flush inside the method for peeking at logs instead of the LogMessage() function
The text was updated successfully, but these errors were encountered: