TagStorage can get corrupted during writing

[kirillt]

Sometimes, the file serving as tag storage becomes corrupted:

...
117421-975209069:visa
92126-3414598226:hmm
60272-1305102029:music
34362-6340143
(END OF FILE)

In the example above, resource id isn't complete and no tags following.

In fact, there is even bigger problem: in such cases, half of the storage is lost. Thanks to backup mechanism the loss can be mitigated. However, this must not happen at all.

Atomic writing should be implemented.

[kirillt]

We need to explore possibilities to prevent "dirty writing". Atomic writing could be implemented like this:

1. Create temporary file with unique name (use standard Java/Kotlin methods to obtain it).
2. When writing is done, copy the temporary file into the storage file (.ark/tags in case of tags storage).
   We also must ensure that no storage is written from several threads/coroutines simultaneously. Otherwise, only one of writes would succeed.

There could be other ways to implement this OR ready-made libraries.

• Write files atomically arklib#48

[shubertm]

persist may complete before setTags completes here https://github.com/ARK-Builders/ARK-Navigator/blob/56e01cbc32a79fd593e32b8c6fb7df5a8ad18fef/app/src/main/java/space/taran/arknavigator/mvp/model/repo/tags/PlainTagsStorage.kt#L106-L111
leading to incomplete write to file

[kirillt]

@ShubertMunthali what can cause this? It seems to me that setTags finishes before persist is called..

[shubertm]

We have two coroutines there, setTags is called in Dispatchers.IO scope while persist is called in launch which inherits the same scope. Is it really necessary to spawn another coroutine in this case?

[kirillt]

Superseded by #69