Dear Robert,
I know you wanted to proceed to making new developments in an ASDF 3.2, but I believe the bad default cache computation on Windows makes it a good idea to release an ASDF 3.1.6 sometime soon. Bonus will be some concurrency fix and a better run-program on ECL.
I still advocate merging in the minimakefile branch before 3.1.6. It will really make releasing and testing easier, especially on Windows.
—♯ƒ • François-René ÐVB Rideau •Reflection&Cybernethics• http://fare.tunes.org You may easily play a joke on a man who likes to argue — agree with him. — Edgar Waston Howe
On 8/30/15 Aug 30 -12:54 PM, Faré wrote:
Dear Robert,
I know you wanted to proceed to making new developments in an ASDF 3.2, but I believe the bad default cache computation on Windows makes it a good idea to release an ASDF 3.1.6 sometime soon. Bonus will be some concurrency fix and a better run-program on ECL.
I still advocate merging in the minimakefile branch before 3.1.6. It will really make releasing and testing easier, especially on Windows.
Should we see about making sure we have a working ASDF on ACL for 3.1.6?
I'm wondering if we should back out the change to system pathnames for ticket 1485276. Is that a bug fix, so ok to shove out in 3.1.6, or is it a new feature, which might better be held for 3.2?
I'll see how minimakefile behaves for me and get back to you...
Best, r
Should we see about making sure we have a working ASDF on ACL for 3.1.6?
Yes of course. If we were quick, we'd manage to release just before ACL is finalized so they can ship with it. Not sure we're that quick.
I'm wondering if we should back out the change to system pathnames for ticket 1485276. Is that a bug fix, so ok to shove out in 3.1.6, or is it a new feature, which might better be held for 3.2?
It's between the two. It's also backward compatible with systems that don't use :pathname for their defsystem. Now 1- this is the case of all 6 systems in Quicklisp that use package-inferred-system 2- there is no imaginable reason why anyone would currently use :pathname in a package-inferred-system, since it had so far no effect whatsoever
NB: OK, technically, if you combine being a package-inferred-system with having a traditional list of :components, then it did affect the pathname of the components, but no one does that, and it's perverse anyway because it leads to files having two different names — only ASDF itself kind of puns file names this way, for the sake of bootstrapping, and it doesn't use :pathname.
I'd keep that "fix" in 3.1.6, and keep disrecommending its use without a :depends-on ((:version "asdf" "3.1.6")).
I'll see how minimakefile behaves for me and get back to you...
Thanks a lot!
—♯ƒ • François-René ÐVB Rideau •Reflection&Cybernethics• http://fare.tunes.org — Question authority! — Yeah, says who?
-
Faré -
Robert Goldman