Dissolve editlevels.h

[pippijn]

We should get rid of editlevels.h by deleting the editlevels one by one and recording them in the git history along with the description in editlevels.h.

[pippijn]

@pbroadbery you have removed some editlevels. Is there any reason not to remove all of them? Are there any editlevels that should stay there? I think there is the hashcode thing that breaks axiom compatibility. Is that relevant, anymore, or can the old hashcode go away?

[hemmecke]

On 07/28/2013 01:05 AM, Pippijn van Steenhoven wrote:

@pbroadbery you have removed some editlevels. Is there any reason not to remove all of them? Are there any editlevels that should stay there? I think there is the hashcode thing that breaks axiom compatibility. Is that relevant, anymore, or can the old hashcode go away?

---

Reply to this email directly or view it on GitHub:
#21 (comment)

Please be careful with the hashcode thing. FriCAS and Aldor should
coordinate here.

I remember that some years ago there were problems with switching some
edit_level on or off. Don't remember which one though. That must be
somewhere in the aldor or fricas mailing list.

Ralf

[pbroadbery]

On Sun, Jul 28, 2013 at 12:05 AM, Pippijn van Steenhoven <
[email protected]> wrote:

@pbroadbery https://github.com/pbroadbery you have removed some
editlevels. Is there any reason not to remove all of them? Are there any
editlevels that should stay there? I think there is the hashcode thing that
breaks axiom compatibility. Is that relevant, anymore, or can the old
hashcode go away?

No reason, aside from wanting to be cautious about a single big change.
Plus each one will take a 5 minute round trip build (say).. which is a fair
chunk of time. If one of them is required for axiom, then it should be
folded in as a compiler option (or fixed in favour of axiom compatibility
for preference).

Peter

[hemmecke]

https://github.com/hemmecke/fricas-svn/blob/master/src/interp/hashcode.boot#L83
Maybe this is relevant.

[pippijn]

That is a different hashcode. By the way, the aldor version of that doesn't have the prime modulus.

The new hashcode was enabled by default, so it still is, but now the compiler option -Wold-hcodes makes the compiler act as if the editlevel about hashcodes was disabled.

Perhaps the option should be renamed to -Waxiom-hcodes or something.