Re: Polished FileSystem API proposal

Glad to see this proposal has a new draft.

On Sat, Jul 13, 2013 at 9:31 AM, Jonas Sicking <jonas@sicking.cc> wrote:

> Executive Summary (aka TL;DR):
> Below is the mozilla proposal for a simplified filesystem API. It
> contains two new abstractions, a Directory object which allows
> manipulating files and directories within it, and a FileHandle object
> which allows holding an exclusive lock on a file while performing
> multiple read/write operations on it.
>
> It's largely modeled after posix, but because we've tried to keep it
> author friendly despite it's asynchronous nature, it differs in a few
> cases.
>
> There are opportunities for further simplifications by straying
> further from posix. It's unclear if this is desired or not.
>
> Detailed proposal:
>
> partial interface Navigator {
>   Promise<Directory> getFilesystem(optional FilesystemParameters
> parameters);
> };
>
> interface Directory {
>   readonly attribute DOMString name;
>

So neither File nor Directory has 'path' attribute but only exposes 'name'.
It feels a bit inconvenient but is it intentional?

File object is meant to be a snapshot and becomes invalid after
modification is made.
Will Directory follow the same model?
Say, if a Directory is acquired then moved (renamed) to another name, does
the Directory object keep functioning or become invalid?


>   Promise<File> createFile(DOMString path, MakeFileOptions options);
>   Promise<Directory> createDirectory(DOMString path);
>
>   Promise<(File or Directory)> get(DOMString path);
>
>   Promise<void> move((DOMString or File or Directory) entry,
>                      (DOMString or Directory or DestinationDict) dest);
>   Promise<void> copy((DOMString or File or Directory) entry,
>                      (DOMString or Directory or DestinationDict) dest);
>   Promise<boolean> remove((DOMString or File or Directory) path,
>                        optional DeleteMode recursive = "nonrecursive");
>

One of the major requests we've gotten in Chrome's FS API for potentially
long-running copy/remove operations is the ability to get progress and to
abort the operation.
Does it make sense to make them return AbortableProgressPromise when they
run recursively?

It's possible that an app creates tons of files in a directory and then
copies the entire directory to another, which could take long time
(depending on implementation).

(Also +1 to have a separate method for recursiveRemove or removeDepp)

  Promise<FileHandle> openRead((DOMString or File) file);
>   Promise<FileHandleWritable> openWrite((DOMString or File) file,
>         optional CreateMode createMode = "createifneeded");
>   Promise<FileHandleWritable> openAppend((DOMString or File) file,
>         optional CreateMode createMode = "createifneeded");
>
>   EventStream<(File or Directory)> enumerate();
>   EventStream<File> enumerateDeep();
>

Can this enumeration/stream be stopped halfway?  If a directory contains
thousands of files callers may not want to keep the disk spinning until the
enumeration reaches the end.

};
>
> interface FileHandle
> {
>   readonly attribute FileOpenMode mode;
>   readonly attribute boolean active;
>
>   attribute long long? location;
>
>   Promise<File> getFile();
>   AbortableProgressPromise<ArrayBuffer> read(unsigned long long size);
>   AbortableProgressPromise<DOMString> readText(unsigned long long
> size, optional DOMString encoding = "utf-8");
>
>   void abort();
> };
>
> interface FileHandleWritable : FileHandle
> {
>   AbortableProgressPromise<void> write((DOMString or ArrayBuffer or
> ArrayBufferView or Blob) value);
>
>   Promise<void> setSize(optional unsigned long long size);
>
>   Promise<void> flush();
> };
>
> partial interface URL {
>   static DOMString? getPersistentURL(File file);
> }
>
> // WebIDL cruft that's largely transparent
> enum PersistenceType { "temporary", "persistent" };
> dictionary FilesystemParameters {
>   PersistenceType storage = "temporary";
> };
>
> dictionary MakeFileOptions {
>   boolean overwriteIfExists = false;
>

The term 'overwrite' feels a bit confusing as it could mean 'overwriting a
file without truncating'.
If what it really means is truncate-if-exists can we rename this to
'truncateIfExists' (or { ifExists: "truncate" }) so that it becomes clearer
it works just like O_TRUNC?

  (DOMString or Blob or ArrayBuffer or ArrayBufferView) data;
> };
>
> enum CreateMode { "createifneeded", "dontcreate" }
> enum DeleteMode { "recursive", "nonrecursive" }
>
> dictionary DestinationDict {
>   Directory dir;
>   DOMString name;
> };
>
> enum FileOpenMode { "read", "write", "append" };
>
> So this API introduces 2 classes: Directory and FileHandle. Directory
> allows manipulation of the files and directories stored inside that
> directory. FileHandle represents an exclusively opened file and allows
> manipulation of the file contents.
>
> The behavior is hopefully mostly obvious. A few general comments:
>
> The functions on Directory that accept DOMString arguments for
> filenames allow names like "path/to/file". If the function creates a
> file, then it creates the intermediate directories. Such paths are
> always interpreted as relative to the directory itself, never relative
> to the root.
>
> We were thinking of *not* allowing paths that walk up the directory
> tree. So paths like "../foo", "..", "/foo/bar" or "foo/../bar" are not
> allowed. This to keep things simple and avoid security issues for the
> page.
>
> Likewise, passing a File object to an operation of Directory where the
> File object isn't contained in that directory or its descendents also
> results in an error.
>
> One thing that is probably not obvious is how the FileHandle.location
> attribute works. This attribute is used by the read/readText/write
> functions to select where the read or write operation starts. When
> .read is called, it uses the current value of .location to determine
> where the reading starts. It then fires off an asynchronous read
> operation. It finally synchronously increases .location by the amount
> of the 'size' argument before returning. Same thing for .write() and
> .readText().
>
> This means that the caller can simply set .location and then fire off
> multiple read or write operations which automatically will happen
> staggered in the file. It also means that the caller can set the
> location for next operation by simply setting .location, or can check
> the current location by simply getting .location.
>
> Setting .location to null means "go to the end".
>
> Note that getting or setting .location does not need to synchronously
> call seek, or do any IO operations, in the implementation. Instead the
> implementation simply tracks .location in the API implementation.
> Whenever a read or write operation is scheduled, the current .location
> is sent along with the operation information to the IO thread and the
> seek can happen there. Many times the implementation can optimize out
> the seek entirely.
>
> The FileHandle class automatically closes itself as soon as the page
> stops posting further calls to .read/.readBinary/.write to it. This
> happens once the last Promise returned from one of those operations
> has been resolved, without further calls to .read/.readBinary/.write
> having happened.


Can the same FileHandle be reused after it's closed, or one need to create
a new FileHandler to start another sequence of read/write?


> This is similar to IDB transactions, though obviously
> there are no transactional semantics here. I.e. there is no way to
> roll back any changes.
>
> There are a few things that we did have disagreements on and which
> would be worth debating:
>
> Is the setup around the FileHandle.location attribute a good idea?
> Some people found it confusingly different from posix.
>
> There's a few more "mode" flags in various functions than I like. In
> particular the "recursive" flag for Directory.remove was debated. Do
> we really need the ability to call .remove on a directory and have it
> fail if the directory isn't empty? And should it really be the default
> behavior?
>
> Likewise, can we get rid of the "createifneeded" vs. "dontcreate"
> switch for .openWrite()/.openAppend()?
>
> What about the overwriteIfExists flag for createFile?
>

I think I like the idea { ifExists: foo } for createFile and { ifNotExists:
bar } for openRead/Write better.



> Do we really need the .openAppend() function? Or is it ok to ask
> people to use .openWrite() and then go to the end before writing?
>
> Finally, there was debate about if we need a Directory abstraction at
> all, or if we could create something simpler which relied on string
> munging instead.
>
> Some examples of what code would look like:
>
> // Save some downloaded data into a new file:
> navigator.getFilesystem().then(function(root) {
>   root.createFile("myfile.txt", { data: xhr.response });
> });
>
> // Append 5 bytes to the end of a large existing file:
> navigator.getFilesystem().then(function(root) {
>   return root.openAppend("largefile.dat");
> }).then(function(handle) {
>   return handle.write(new Uint8Array([1, 1, 2, 3, 5]));
> });
>
> // Increase the 100th byte in large existing file:
> var fileHandle;
> navigator.getFilesystem().then(function(root) {
>   return root.openAppend("dir/highscores");
> }).then(function(handle) {
>   fileHandle = handle;
>   fileHandle.location = 100;
>   return handle.read(1);
> }).then(function(buffer) {
>   assert(buffer.byteLength === 1);
>   var view = new Uint8Array(buffer);
>   view[0]++;
>   fileHandle.location--;
>   return handle.write(buffer);
> });
>
> I hope to send this proposal to public-script-coord soon after some
> debate on this list.
>
> [1]
> http://lists.w3.org/Archives/Public/public-webapps/2013AprJun/0382.html
>
> / Jonas
>
>