711841 - Add VoidString() or NullString() or NullDOMString()

Status: RESOLVED FIXED

(Core :: XPCOM, defect)

Description

[Olli Pettay [:smaug][[e-mail]]]

Atm, one needs to explicitly call SetIsVoid(). Would be nice to have something
close to EmptyString()

Comment 1

[Olli Pettay [:smaug][[e-mail]]]

Something like
#define NullString() nsXPIDLString()
#define NullCString() nsXPIDLCString()

Comment 2

[Olli Pettay [:smaug][[e-mail]]]

Or better to implement it like Empty(C)String

Comment 3

[Olli Pettay [:smaug][[e-mail]]]

Attachment: something like this

Since nsXPIDLString has the behavior anyway, we should be able to reuse that.

https://tbpl.mozilla.org/?tree=Try&rev=ec49c8378fd2

Comment 4

[Boris Zbarsky [:bzbarsky]]

SetDOMStringToNull exists.

Comment 5

[Olli Pettay [:smaug][[e-mail]]]

This is not about a method which sets a string to null, but having a null string, which can be easily passed as a parameter to methods, or used in ctor member initialization.

Comment 6

[Benjamin Smedberg]

Comment on attachment 582680 [details] [diff] [review]
something like this

glandium, will this cause a static ctor to run, and do you think that's a profile? Otherwise I think this is fine.

Comment 7

[Mike Hommey [:glandium]]

The patch as is won't cause a static ctor, but will cause a mutex being used each time NullString and NullCString are used.

Comment 8

[Olli Pettay [:smaug][[e-mail]]]

IRC:
glandium >	smaug: mmm actually looking at the code gcc generates, it only acquires the lock if it the constant wasn't already initialized, so it's not as bad as i suggested

That doesn't sound bad to me, and EmptyString has the same behavior.

Comment 9

[Mike Hommey [:glandium]]

Comment on attachment 582680 [details] [diff] [review]
something like this

rs=me

Comment 10

[Olli Pettay [:smaug][[e-mail]]]

https://hg.mozilla.org/mozilla-central/rev/3328809425d2