Global variables in DSC

[gaow]

In this ticket #70 (comment) I showed the use of DSC global variable:

DSC:
  variable:
     some_var:  '...'

I've agreed with @stephens999 that we still would use $ symbol instead of $() to access it, because it is similar to pipeline variable in scope. However the difference here is that these variables are constants.

There are two points to discuss:

1. Do we need these global variables? I vote for Yes because I'm sure there will be shared parameters among all modules, for things such as path and computational environment configurations (we've not talked about execution on remote computers eg RCC).
2. How should they be distinguished? We are not forcing any conventions on these variables (eg all capital letters) and we'll check and raise error if there is a name conflict with pipeline variables. Is it good enough?

[stephens999]

are these constant variables? so can't be output?

[gaow]

Yes these are constants. User cases I can think of are typically strings such as file names, paths, computational environment settings (walltime, mem, etc -- though we'll have global defaults in separate config file just like snakemake), or a particularly meaningful choice of sets of parameters that we want to treat as constants.

[stephens999]

i'm good with these. How about global: instead of variable: as keyword?

[stephens999]

and $some_var instead of some_var ?

[gaow]

Agreed!

But what is the behavior when there is a conflict (my question 2 above) with pipeline variables? Report an error or warning? I guess an error? If you think the same then we're good to close hte issue.

[stephens999]

yes, I agree, global variables must have different names than pipeline variables.

…

On Tue, May 9, 2017 at 9:10 PM, gaow ***@***.***> wrote: Agreed! But what is the behavior when there is a conflict (my question 2 above) with pipeline variables? Report an error or warning? I guess an error? If you think the same then we're good to close hte issue. — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#81 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/ABt4xfnoYLdZnLfgtX8qlNZbBckGVCLwks5r4RymgaJpZM4NSZ_f> .

[stephens999]

... and can't be outputs from module? On Wed, May 10, 2017 at 9:36 AM, Matthew Stephens <stephens999@gmail.com> wrote:

…

yes, I agree, global variables must have different names than pipeline variables. On Tue, May 9, 2017 at 9:10 PM, gaow ***@***.***> wrote: > Agreed! > > But what is the behavior when there is a conflict (my question 2 above) > with pipeline variables? Report an error or warning? I guess an error? If > you think the same then we're good to close hte issue. > > — > You are receiving this because you were mentioned. > Reply to this email directly, view it on GitHub > <#81 (comment)>, > or mute the thread > <https://github.com/notifications/unsubscribe-auth/ABt4xfnoYLdZnLfgtX8qlNZbBckGVCLwks5r4RymgaJpZM4NSZ_f> > . >

[gaow]

Yes any pipeline variable will result in a conflict. I'll ensure that's checked for. Closing this issue now.