www.digitalmars.com         C & C++   DMDScript  

digitalmars.D - Deprecate -v1 ?

reply Lutger <lutger.blijdestijn gmail.com> writes:
With the recent branch in compilers for D1 and D2, it looks to me that 
there are now at least three versions of D:

D version 1 as (sort of) defined by the -v1 switch
D1 stable branch
D2 branch

Furthermore, I've seen comments that dmd 1.014 introduced some 
incompatibilities with previous releases, and therefor a compiler 
version between dmd 1.0 and dmd 1.013 was used.

Because the branching is essentially an altenative for the -v1 switch, 
the D1 branch is likely to remain stable and receive bug fixes, doesn't 
it make sense to have only one version of D1 and thus deprecate D -v1?

There may be other solutions, but the difference between D1 with and 
without -v1 may be confusing. What do you think?
Jun 19 2007
parent Sean Kelly <sean f4.ca> writes:
Lutger wrote:
 With the recent branch in compilers for D1 and D2, it looks to me that 
 there are now at least three versions of D:
 
 D version 1 as (sort of) defined by the -v1 switch
 D1 stable branch
 D2 branch
 
 Furthermore, I've seen comments that dmd 1.014 introduced some 
 incompatibilities with previous releases, and therefor a compiler 
 version between dmd 1.0 and dmd 1.013 was used.

Header generation in 1.014 and beyond generates "ref" where "inout" was generated previously. That caused problems for a while. Also, I suppose it's worth mentioning that 1.015 and beyond have TypeInfo for "const" and "invariant", which suggests that either some of the const features are in D 1.0, or the addition was a mistake. These classes weren't added to Tango, but they are in Phobos.
 Because the branching is essentially an altenative for the -v1 switch, 
 the D1 branch is likely to remain stable and receive bug fixes, doesn't 
 it make sense to have only one version of D1 and thus deprecate D -v1?

It does to me. Sean
Jun 19 2007