Model Crashes When OUTPUT_FORCE=TRUE

[jhamman]

When using OUTPUT_FORCE = TRUE, the model crashes with a segmentation fault: bad access.

Looking into this a bit deeper, this problem is coming in two locations where access to the veg_con structure, isn't protected by a if (!options.OUTPUT_FORCE) {} block. This occurs in two blocks of the initialize_atmos() code:

1. Assign N_ELEM for veg-dependent forcings,
2. Albedo, LAI, and Fractional Veg Cover.

I've already applied a fix locally but this is probably going to mean we release a hotfix/4.2.b tag.

[bartnijssen]

@jhamman - I am confused about the tag naming. I thought bug fixes were identified by letters (4.2.a, 4.2.b) rather than numbers. I already didn't understand 4.2.1, but am concerned that this will cause confusion for the 4.X series. Even if going forward it'll be <major release>.<minor release>.<bug fix> then that would be easier from 5.0 going forward (i.e. 5.0.1 is a bug fix for 5.0). I'd prefer if we retag the 4.2.1 tag as 4.2.a and call this one 4.2.b, given that the 4.1.0 to 4.1.2` were essentially minor releases rather than bug fixes.

[jhamman]

@bartnijssen - We stopped using the alphabetic bugfix identifier when we went to 4.2.0. That discussion was in issue #49. When the issues was close, I specifically addressed the convention that was used for the 4.2.1 release.

[bartnijssen]

I know we had that discussion. After seeing it in practice, I really don't like it because it is misleading within the 4.X universe. For this reason I am making the decision to change it and adopt it from 5.X on instead.

On Jan 22, 2015, at 9:56 AM, Joe Hamman notifications@github.com wrote:

@bartnijssen - We stopped using the alphabetic bugfix identifier when we went to 4.2.0. That discussion was in issue #49. When the issues was close, I specifically addressed the convention that was used for the 4.2.1 release.

—
Reply to this email directly or view it on GitHub.