Thoughts on Backward Compatibility

cc cc at nevernet.com
Tue Feb 20 13:02:52 UTC 2024


On Friday, 16 February 2024 at 04:38:03 UTC, H. S. Teoh wrote:
> GC.  Instead of bending over backwards trying to woo the @nogc 
> crowd

I just want to chime in here:  I'm a GC minimalist who avoids it 
whenever reasonably possible and only uses it when it makes the 
most sense.  And I really don't care about @nogc.  Feel free to 
dump it and work on cooler stuff, you have my support!

> trying to woo an amorphous group of hypothetical potential users

I feel like this describes a lot of the general philosophy I get 
observing these forums lately.  A perhaps overly cynical 
interpretation would be that (some of?) the D community is so 
insecure about losing any more of its already well-known to be 
small userbase that it's terrified to make any serious meaningful 
positive changes in the event some unknown person somewhere with 
a dub package that hasn't been touched in 7 years gets annoyed 
typing build and decides to move on (hey, I get annoyed typing 
dub build *every time*).  Or some vague future new user to whom 
this will happen 7 years hence.  Every time I come to General I 
see another thread with a deep, introspective, heavily passionate 
argument about why we can't have nice thing because of some 
astronomically remote edge case and everything gets frozen into a 
moebius loop of trying to figure out how to account for every 
possible combinatorial ways it might be used or misused.  I'll 
grant lack of foresight has long been the C Family Curse, but 
there is such a thing as being too navel-gazing as well.

Every time I see one of these hot topics and start drawing up a 
response, I watch it spiral deeper into interdependent debates, 
bringing up every flaw that D, phobos, and C++ have ever 
experienced in their lifetimes, of why it needs to be absolutely 
perfect to some exacting cosmological standard so that the 
nebulous supercorporation that may or may not use it in some 
unspecified future will be sufficiently satisfied with the 
implementors' fidelity, and sigh and delete the post.  I've 
discarded more drafts to this forum than I've ever submitted.  
Why argue with the heavyweights?  They've got all the scientific 
proof that doing anything that may need to change someday is 
simply impermissible, and all I have is a fondness for nice 
things.  Nice things are what drew me to D in the first place, 
but now nice things are anathema, because we might be required to 
take responsibility for them someday.  Why get a dog if you have 
to walk it?


More information about the Digitalmars-d mailing list