On Mon, Mar 17, 2008 at 09:10:18PM +0100, Tobias C. Rittweiler wrote:

Hi,

I'm working having an (INSPECT FOO) form cause the Slime Inspector to pop up (on those implementations that provide a way to hook into the their inspection machinery.)

At the moment, we maintain only one *Slime Inspector* buffer, this has to be changed due to concurrent evaluation in several threads: For instance, think of code invoking INSPECT run at the REPL, and code invoking INSPECT evaluated in a *slime-scratch* buffer.

I can think of two different schemes of maintaining inspector buffers:

a) At the moment, each `C-c I' causes an already existing inspector buffer to be overwritten, change that so that each `C-c I' creates a new buffer. Each invocation of INSPECT also creates a new independent buffer.

b) Each `C-c I' doesn't create a new inspector buffer, but peruses the one created by `C-c I' the last time (or create it.) Each invocation of INSPECT creates a new independent buffer that's not touched by `C-c I'

Which scheme do you prefer? (Or would you rather have it entirely different?)

definitely a) since I often use multiple repls

* Tobias C. Rittweiler [2008-03-17 21:10+0100] writes:

Hi,

I'm working having an (INSPECT FOO) form cause the Slime Inspector to pop up (on those implementations that provide a way to hook into the their inspection machinery.)

At the moment, we maintain only one *Slime Inspector* buffer, this has to be changed due to concurrent evaluation in several threads: For instance, think of code invoking INSPECT run at the REPL, and code invoking INSPECT evaluated in a *slime-scratch* buffer.

I think threads and inspectors are orthogonal issues. You could have multiple inspectors with one thread but also a single inspector for multiple threads.

I can think of two different schemes of maintaining inspector buffers:

a) At the moment, each `C-c I' causes an already existing inspector buffer to be overwritten, change that so that each `C-c I' creates a new buffer. Each invocation of INSPECT also creates a new independent buffer.

b) Each `C-c I' doesn't create a new inspector buffer, but peruses the one created by `C-c I' the last time (or create it.) Each invocation of INSPECT creates a new independent buffer that's not touched by `C-c I'

Which scheme do you prefer? (Or would you rather have it entirely different?)

I think INSPECT should do the same as C-c I. Two commands would also also be reasonable, one command to create a new inspector and the other to reuse the current inspector. We could also us a prefix-arg to control whether a new inspector should be created.

Helmut.