Python-clu

Other

* Bump version: 0.2.7 → 0.2.8. [Alexander Böhn]

* Rewrote “ls(…)” and “ls_la(…)” from “clu.fs.filesystem.Directory” ... to use the new less-cumbersomely-verbose modes of dealing with “clu.fs.misc.suffix_searcher(…)” [Alexander Böhn]

* Rewrote a lot of “clu.fs.filesystem.Directory.flatten(…)” ... as well as rewrites in “clu.fs.misc.suffix_searcher(…)” and “clu.testing.utils.countfiles(…)” – the latter of which now also takes a “suffix” argument to only count files matching a specific suffix, like duh ... expanded the “flatten(…)” method tests to cover all the new combos of using the suffix-related arguments and double-checked the output of everything and whatnot ... ALSO FIXED MANY BUUUUUUUGGS. [Alexander Böhn]

* Exporter checks in “clu.testing.pytest.clumods” and “show-modules.py” ... beforehand, there was a hardcoded list of “CLU modules” in a static tuple in “clu.constants.data” which really was a list of modules in CLU that used the “clu.exporting.Exporter” class to export their shit. ... now there still is; the difference is that that tuple can now contain any valid module in CLU and the two places where the tuple gets iterated also check to see if the module they’re exporting contains an exporter – if it doesn’t, whatever that bit of code returns won’t contain said module ... clearly this is a janky system but we currently need it to test that the Exporter registration system works in the first place; it could get pared down to like only being used in one or two instances, I feel like. [Alexander Böhn]

* Changed “followlinks=True” in “Directory.walk(…)” arguments ... plus some random updates and minutiae in “clu.fs.filesystem” and “clu.testing.utils” [Alexander Böhn]

* Confirming the counts of the flattened directory contents ... using a “countfiles(…)” function, which in itself isn’t really any type of anything – but its existence did necessitate the creation of a new “clu.testing.utils” module. [Alexander Böhn]

Other

* Bump version: 0.2.6 → 0.2.7. [Alexander Böhn]

* I think it’s irritating how .jpg and .jpeg are valid JPEG suffixes. [Alexander Böhn]

Other

* Bump version: 0.2.5 → 0.2.6. [Alexander Böhn]

* We now have a “Directory.flatten(…)” instance method ... plus a working test stub, plus helper methods (one of which let us rewrite some of “Directory.zip_archive(…)” to omit inlining the “relparent(…)” lambda). I wrote all of this up at the bar while standing up and drinking whiskey talking intermittently to other patrons so I am calling this effort NOT BAD AT ALL. [Alexander Böhn]

* Makefile rules for running pytest and tox ... I mean, it’s cool and all that the pytest fixture stuff is now a “plugin” module, with a setuptools endpoint and shit, instead of just a conftest.py file (which was working just fine by the way) but pytest, while a nice program and all, completely loses its fucking shit completely under certain conditions – one of which is, if somewhere or somehow during its utterly Byzantine loading/discovery/setup/config phase it is told to load one of these “plugin” modules more more than exactly once, it crashes like a fucking trainwreck and spews out what has to literally be the largest and most illegible traceback I have ever seen (and I worked with JSP and Servlets 1.0 in the fucking late 1990s). ... Anyway. So pytest isn’t all that bad aside from one or two of these occasional supernova-class exception belches every so often – once I sifted through the wreckage for the offending file I found the answer reading the pytest code, which was completely decent: legible, full of well-written comments and docstrings; aside from the plague of leading underscores that infects a lot of Pythonilinguists I thought it was great. So yeah I read it, figured out the fix (which itself wasn’t anything too nasty or circuitous) and that’s that. ... So yeah that’s the reason for this long-winded commit note: it’s so easy to talk shit about code and be like MOTHERFUCKER WHAT IN THE NAME OF LOVECRAFTIAN TWAT IS THIS AWFULNESS, WHO EVEN THINKS LIKE THAT and soforth; but so it’s necessary to point out when shit is not bad, and especially when shit is in fact somehow good. So yeah GOOD ON YOU, PYTEST, YOU GET A COOKIE!!! Indeed. [Alexander Böhn]

* The “clu.fs.filesystem.script_path()” function is poorly behaved ... particularly in normal sdist installs. Its test code has been branded with the shameful and dreaded X-FAIL for the moment. [Alexander Böhn]

Other

* Bump version: 0.2.4 → 0.2.5. [Alexander Böhn]

* Getting rid of root-level conftest.py, in favor of “clu.testing” [Alexander Böhn]

* A docstring! A docstring for this function straight away!!! ... if you say it in like a King Arthur voice it’s kinda funny. [Alexander Böhn]

* This is evidently how console-script endpoints should work. [Alexander Böhn]

Add

* Added a stupid little version-string script. [Alexander Böhn]

Other

* Bump version: 0.2.3 → 0.2.4. [Alexander Böhn]

* Made the stupid little version-string script a setuptools entrypoint. [Alexander Böhn]

Other

* Bump version: 0.2.2 → 0.2.3. [Alexander Böhn]

* Moving the pytest fixtures that use CLU formally into the project. [Alexander Böhn]