Rcompression To Do list

[Done] names on the zip archive (@elements) becoming numbers when duplicates. Are we taking the rownames when we have the actual names!

  content = I(c(myFile = "This is raw text",
                  otherFile = paste(letters, collapse = "\n"),
                  x = "a", x = "b", x = "c"))

  zip("asis.zip", content, append = FALSE)
  z = zipArchive("asis.zip")      

When rewriting a zip file, want to preserve times. So need to be able to specify these as a vector to zip().

For RdzArchive files, should .toc not count when indexing by integer.

i.e. x[[1]] gives the second element since .toc is first.

[Done] Get the last element named 'whatever'. Do this for [[ and $ operators on ZipArchive.

Done for [[ in RdzArchive.
[Done] Move to ZipArchive.

Add ability to remove an entry.

Load them all up, remove one or more, then zip them back. For Rdz files need to avoid deserializing first.

Getting problems about NULL value passed to R_getNativeReference.

Is this just with remove entries - zipRemoveEntries?

When update a RdzArchive, the .toc is out of date in the file.

Not much we can do about this unless we add a new one.
When create the object in zipArchive() we can cache the Sys.time() and compare that to the file.info()[1, "mtime"] to see if the file has been modified since the elements were cached.

[Done] When we have several elements with the same name in an archive, allow accessing the last one?

ar[["x", last = TRUE]] # default ar[["x", last = FALSE]]

Global comment.

Use zipOpen2 in minizip rather than zipOpen().

[Done] When we update a ZipArchive, the info slot is out of date

We can add a row to the date frame when we append.
[Done] Updated now, but fix the row names (now .toc, .toc1, .toc2)

[Done] getZipInfo( comment = TRUE) causing problems

Returning garbage. Initialization problem in C.

[Done] setting rownames on the getZipInfo() can cause problems because of duplicate names.

library(Rcompression)
x = I(c(a = "ABC", b = "DEF", c = "GHIJKLML"))
zip("foo.zip", x, append = FALSE)
zip("foo.zip", x)
z = zipArchive("foo.zip")
getZipInfo(z)      

[Done] Allow comments on the zip elements. And also for the entire zip file, i.e. a global comment.

Done for the elements, but how do we see them? getZipInfo() now optionally returns them.
But they are truncated.
Seems okay now. See tests/comments.R

Reading a file from a zip file seems to take a long time. Are we reading the whole file each time?

[Done] Encoding issues in tests/zip.R

Don't think so. Just have to supply the password. (Was that the problem?)

Add the deflate functions for gzip and bzip2.

gzip done. But we cannot gunzip() the result.
Still true?

gunzip(gzip("This is a string"))      

No bzip2 facilities.

Allow the user to provide a connection on which to provide the compressed input so that the routines pull the data from that rather than having it in a buffer ahead of time.

Hookup to the RCurl write_binary_data routine to have inflation be done on the fly.

[Done] If given I(c("A", "B")), preserve the I across the elements when making a list.

In the tests now.

[Done] On Windows, can we have, e.g., c:/ in the name of a directory. What about \

Yes, but the regular unzip replaces c:/ with c_/, i.e. we get a top-level directory named c_

[Done] Check directories on Windows

content = list( "A/myFile" = I("This is raw text"), "B/otherFile" = I(paste(letters, collapse = "\n")), top = I("A string"))
zip("/tmp/dir.zip", content)      

Then unzip externally to see if the zip facilities do the right thing.

[Done] Check directories for zip, i.e. when we use hierarchical names

[Done] Put the current date for in-memory elements in call to zip().

[Done for []. ]Make the x[] and x[c(a, b, c)] read sequentially rather than starting over each time.

Could find a, b, c and figure out the order.