removing .elc files for `make bootstrap'

removing .elc files for `make bootstrap'

[Miles Bader]

Any objection to me changing `make bootstrap' to remove all .elc files
at the beginning?  Their presence can cover up bugs[*], and as they all
get recompiled by make bootstrap anyway, initially removing them
shouldn't appreciably lengthen the time required for the bootstrap.

[*] See the recent emacs-devel thread `Subject: batch-bytecomp backtrace?'

Thanks,

-Miles
-- 
"I distrust a research person who is always obviously busy on a task."
   --Robert Frosch, VP, GM Research

Re: removing .elc files for `make bootstrap'

[Juanma Barranquero]

On 19 Apr 2004 07:42:01 +0900, Miles Bader <miles@gnu.org> wrote:

> Any objection to me changing `make bootstrap' to remove all .elc files
> at the beginning?

FWIW, that's what the Windows makefiles already do.

                                                           /L/e/k/t/u

Re: removing .elc files for `make bootstrap'

[Kenichi Handa]

In article <87d664nc1y.fsf@tc-1-100.kawasaki.gol.ne.jp>, Miles Bader <miles@gnu.org> writes:

> Any objection to me changing `make bootstrap' to remove all .elc files
> at the beginning?  Their presence can cover up bugs[*], and as they all
> get recompiled by make bootstrap anyway, initially removing them
> shouldn't appreciably lengthen the time required for the bootstrap.

> [*] See the recent emacs-devel thread `Subject: batch-bytecomp backtrace?'

I agree with that change.  Just recently I was bitten by
this problem when I synchronized emacs-unicode-2 to the
trunk.

---
Ken'ichi HANDA
handa@m17n.org

Re: removing .elc files for `make bootstrap'

[Jérôme Marant]

Quoting Miles Bader <miles@gnu.org>:

> Any objection to me changing `make bootstrap' to remove all .elc files
> at the beginning?  Their presence can cover up bugs[*], and as they all
> get recompiled by make bootstrap anyway, initially removing them
> shouldn't appreciably lengthen the time required for the bootstrap.

Couldn't the bootstrap target simply depend on the maintainer-clean one?

-- 
Jérôme Marant

Re: removing .elc files for `make bootstrap'

[Stefan Monnier]

> Any objection to me changing `make bootstrap' to remove all .elc files
> at the beginning?  Their presence can cover up bugs[*], and as they all
> get recompiled by make bootstrap anyway, initially removing them
> shouldn't appreciably lengthen the time required for the bootstrap.

I'd rather insist on people using

    make maintainer-clean bootstrap

so we can keep working on making `make bootstrap' and `make' closer
to each other.


        Stefan

Re: removing .elc files for `make bootstrap'

[Stefan Monnier]

> Any objection to me changing `make bootstrap' to remove all .elc files
> at the beginning?

I need/want a bootstrap target that is "restartable".
So we should distinguish between `maintainer-clean' which prepares for
a clean boostrap and `bootstrap' which just builds the files assuming that
the dependencies are correct (or that the files were removed by
maintainer-clean if the dependencies were not).


        Stefan

Re: removing .elc files for `make bootstrap'

[Miles Bader]

Stefan Monnier <monnier@iro.umontreal.ca> writes:
> > Any objection to me changing `make bootstrap' to remove all .elc files
> > at the beginning?
> 
> I need/want a bootstrap target that is "restartable".

I'd love that too!

> So we should distinguish between `maintainer-clean' which prepares for
> a clean boostrap and `bootstrap' which just builds the files assuming that
> the dependencies are correct (or that the files were removed by
> maintainer-clean if the dependencies were not).

Why not keep the current semantics for `bootstrap', and add `bootstrap-cont'
or something to continue with a previous bootstrap?

I'm not sure how important it is to maintain compatibility, but it seems
like `make bootstrap' has become something of an idiom -- many people
think of it as what you do when you want to double-check an emacs tree
for buildability.

-Miles
-- 
My spirit felt washed.  With blood.  [Eli Shin, on "The Passion of the Christ"]

Re: removing .elc files for `make bootstrap'

[Kim F. Storm]

Miles Bader <miles@gnu.org> writes:

> Why not keep the current semantics for `bootstrap', and add `bootstrap-cont'
> or something to continue with a previous bootstrap?

make reboot

  :-)

-- 
Kim F. Storm <storm@cua.dk> http://www.cua.dk

Re: removing .elc files for `make bootstrap'

[Kenichi Handa]

In article <jwvn05859nk.fsf-monnier+emacs@gnu.org>, Stefan Monnier <monnier@iro.umontreal.ca> writes:

>>  Any objection to me changing `make bootstrap' to remove all .elc files
>>  at the beginning?

> I need/want a bootstrap target that is "restartable".
> So we should distinguish between `maintainer-clean' which prepares for
> a clean boostrap and `bootstrap' which just builds the files assuming that
> the dependencies are correct (or that the files were removed by
> maintainer-clean if the dependencies were not).

I think the basic problem of having old .elc is that
`require' loads it even if there exists a newer .el.  Can't
we solve this problem by having a flag variable which, if
non-nil, forces loading whichever the newer one.  Then we
can bind that variable to t on bootstrapping (or always on
byte-compiling).

---
Ken'ichi HANDA
handa@m17n.org