Solve outstanding issues

[JJ]

• Why pod names are lower-cased? finanalyst/pod-cached#24 Why pod names are lower-cased?
• Flapper: it fails some times finanalyst/pod-cached#23 Flapper: it fails some times
• Use specific exceptions classes finanalyst/pod-cached#22 Use specific exceptions classes
• Pod::To::Cached apparently hangs up when a new file is added finanalyst/pod-cached#21 Pod::To::Cached apparently hangs up when a new file is added
• Tests for several caches at the same time? finanalyst/pod-cached#20 Tests for several caches at the same time?
• If files are moved and the cache exists, it hangs up finanalyst/pod-cached#18 If files are moved and the cache exists, it hangs up
• The constructor wrongly assumed the existence of a cache if the directory exists finanalyst/pod-cached#17 The constructor wrongly assumed the existence of a cache if the directory exists

[finanalyst]

re. #24, lower cased names.
I lower cased them in order to catch mis-spelt file names.
eg., if case-sensitive, then pod::To::Cached is different to Pod::To::Cached and a mis-spelt name would be allowed into the cache as well as the correct one.

Case sensitivity did not seem as important for documentation files as it might be for variables.

However, emitting a warning when a mis-spelling is detected is perhaps better.

[JJ]

But both paths might coexist in Linux; that means that it will cause a clash for valid path names.

[finanalyst]

The point is that we are dealing with Pod files. Is there any real possibility that documentation files are going to differ by the case of a letter in the name of the file for a valid reason?

Isn't it more likely that the difference of a different case of a letter in a file name is due to a misspelling? This occurred to me several times.

[finanalyst]

I seem to remember there was another reason I added the normalisation to lower case.

Some other package did not work properly when uppercase letters were used.

But I cannot remember exactly what it was.

[finanalyst]

JJ / Antonio, I want to work on the Specific exceptions class issue.
Have you any preference for what it should be called?

[tbrowder]

How does the non-case nature of the Windows file system affect the issue? I've found over the years that Windows users (not devs so much) tend to be careless about such things compared to *nix users.

[JJ]

In general, we try to avoid such confusions, because the same happens to Mac OSX users. But OSX keeps case, and Windows does it too, I think, even if it's mapped to the same file if case is mixed. Anyway, file name in the repository does not have anything to do with real name, it's computed from the key (directory) and something else. We really have to go through hoops to get the original name back in some cases, so I don't really think this really serves some purpose. Also, we do test now the whole stack in Windows, so we'll create specific test cases for this so that we don't find any trouble.

[finanalyst]

The windows exception might be why I normalised to lower case for the cache key names. I also stored the original name of the file in the index.json file so that the case-restored file name could be restored if need be from the key name.
Insisting on case-sensitive file names because *nix can seems to me to be overkill. ???

[JJ]

El lun., 30 sept. 2019 a las 17:52, Richard Hainsworth (< notifications@github.com>) escribió:

The windows exception might be why I normalised to lower case for the cache key names. I also stored the original name of the file in the index.json file so that the case-restored file name could be restored if need be from the key name. Insisting on case-sensitive file names because *nix can seems to me to be overkill. ???

It's not overkill, it's the default value and it's how they are actually stored. It's simply storing the files using the same case that they are stored originally.

[finanalyst]

I'm trying to duplicate the 'moving file' problem.
After running tests 0-040* (not 050*), the test directory has a cache in 't/tmp/ref' and documents in 't/tmp/doc'.
Running perl6 -Ilib -MPod::To::Cached from the root directory of the repo and I enter REPL
I run my Pod::To::Cached $c .= new(:path<t/tmp/rep>)
Then in another terminal I move a Pod file from one sub directory of 't/tmp/doc' to another.
in REPL, I run $c.update-cache and I get no error, nothing hangs or freezes.

I rerun my Pod::To::Cached $c .= new(:path<t/tmp/rep>) Again no ill effects.

I want to create a test file that generates the problem reported.

[JJ]

Just add a test for that. If it does not fail, then the problem might be somewhere else, maybe in Documentable.