Re: Rationale for change in behavior of g_strsplit when passed empty string?

[Darin Adler <darin bentspoon com>]

On Monday, July 16, 2001, at 05:17  PM, Owen Taylor wrote:

> In absence of a clear "right way" we may need to stick to the 1.2
> behavior to avoid breaking existing code in tricky ways.

On Monday, July 16, 2001, at 05:17  PM, Havoc Pennington wrote:

> Fixing this bug sounds reasonable. However, it isn't clear to me that
> the change you describe was intentionally made as part of fixing it.
> Any thoughts on which behavior is more useful or more expected?

I have no real preference here, but I think it would be better to not have 
a change from 1.2 as Owen says. I only noticed this change at all because 
we have tests in eel that started producing incorrect results.

> Finally, using g_strjoin(), g_strsplit() to flatten and unflatten
> vectors is fundementally broken since they don't do escaping.

In the cases I'm dealing with, we are using g_strsplit to do elementary 
parsing of data structures that use a reserved split sequence and have no 
escaping scheme. For example, chewing apart X font names.

I agree completely that more-serious uses require something better. It's 
possible that some of the uses of g_strsplit and g_strjoin in eel and 
nautilus are broken and need some kind of escaping scheme.

    -- Darin