288786 - Add BeginWriting equivalent to frozen string API

Status: RESOLVED FIXED

(Core :: XPCOM, enhancement, P1)

Description

[Benjamin Smedberg]

We need to add an equivalent for BeginWriting to the frozen string API for gecko
1.8.  Darin, if you won't have time for this reassign it back to me.

Comment 1

[Darin Fisher]

Yeah, this is something I really want to do as well.

This means adding NS_C?StringGetMutableData (or something like that) to the
frozen XPCOM exports.

Comment 2

[Darin Fisher]

We may also want a way to set the string buffer length.

Comment 3

[Christian :Biesinger (don't email me, ping me on IRC)]

(In reply to comment #2)
> We may also want a way to set the string buffer length.

Why not make that a single function?

Comment 4

[Darin Fisher]

Yes, that is exactly my plan.  Here's my suggestion:

  PRUint32
  NS_StringGetMutableData(nsAString &aStr,
                          PRUint32 aDataLength,
                          PRUnichar **aData);

The idea here is that you could pass PR_UINT32_MAX as aDataLength to indicate
that you want the current length of the string.  The length of the data buffer
returned by aData is indicated by the return value of this function.  If the
function fails because it could not allocate memory, then aData is set to null,
and the function returns 0 or maybe PR_UINT32_MAX (not sure about what the
function should return if unable to allocate memory).

At any rate, I'm leaning toward this function signature over one that returns
the buffer itself for consistency (somewhat) with NS_StringGetData.

Thoughts?

Comment 5

[Benjamin Smedberg]

Sounds fine, and it should return 0 if allocation was unsuccessful.

Comment 6

[Darin Fisher]

Here's what I have so far:

-----------------------------------------------------------------------------
/**
 * NS_StringGetMutableData
 *
 * This function provides mutable access to a string's internal buffer.  It
 * returns a pointer to an array of characters that may be modified.  The
 * returned pointer remains valid until the string object is passed to some
 * other string function.
 *
 * Optionally, this function may be used to resize the string's internal
 * buffer.  The aDataLength parameter specifies the requested length of the
 * string's internal buffer.  By passing some value other than PR_UINT32_MAX,
 * the caller can request that the buffer be resized to the specified number of
 * characters before returning.  The caller is not responsible for writing
 * null-terminator.
 *
 * @param aStr          abstract string reference
 * @param aDataLength   number of characters to resize the string's internal
 *                      buffer to or PR_UINT32_MAX if no resizing is needed
 * @param aData         out param that upon return holds the address of aStr's
 *                      internal buffer or null if the function failed
 * @return              number of characters or zero if the function failed
 */
NS_STRINGAPI(PRUint32)
NS_StringGetMutableData
  (nsAString &aStr, PRUint32 aDataLength, PRUnichar **aData);
-----------------------------------------------------------------------------

If you see any problems with this, please let me know.

Comment 7

[Darin Fisher]

Attachment: v1 patch

Comment 8

[Christian :Biesinger (don't email me, ping me on IRC)]

Comment on attachment 179628 [details] [diff] [review]
v1 patch

+  return begin.size_forward();

that does not include the null terminator, right?

(hrm, this file has a lot of functions that only differ in data type. could we
use some template functions to avoid code duplication, at least source-wise?
not for this bug, certainly)

Comment 9

[David Baron :dbaron: (⌚️UTC-4, no longer working on Mozilla)]

Comment on attachment 179628 [details] [diff] [review]
v1 patch

>Index: stub/nsXPComStub.cpp
>     &NS_Alloc_P,
>     &NS_Realloc_P,
>     &NS_Free_P,
>     &NS_StringContainerInit2_P,
>-    &NS_CStringContainerInit2_P
>+    &NS_CStringContainerInit2_P,
>+    &NS_StringGetMutableData,
>+    &NS_CStringGetMutableData

Any reason these don't use _P ?

Other than that, sr=dbaron.

Comment 10

[Darin Fisher]

whoops.. yes!  thank you for catching that mistake.

Comment 11

[Darin Fisher]

Comment on attachment 179628 [details] [diff] [review]
v1 patch

This is a new API that would be nice to have in the developer preview.

Comment 12

[Asa Dotzler [:asa]]

Comment on attachment 179628 [details] [diff] [review]
v1 patch

a=asa

Comment 13

[Darin Fisher]

fixed-on-trunk

Comment 14

[Darin Fisher]

*** Bug 242577 has been marked as a duplicate of this bug. ***