Making images in the modern world

When I last used LW really seriously (between 1999 and 2004 I think) I had a hairy infrastructure for making and saving images with code preloaded.  I think the purpose of this was to make loading time better, especially in the presence of lots of patches and possibly large preloaded systems.

Is there any real purpose in doing this now, as opposed to having a combination of init files which load things and set logical pathnames and (for instance) quicklisp to drag in third-party stuff on demand?

My intuition is that it's not (other than for the special case of making a console image) because things are now so fast.  But as I'm running 6.1 I don't have any significant number (or, in fact, any) patches to load at startup and, so far, not very much code (though I am loading Quicklisp, and that seems pretty zoomy on my 2008 macbook): if anyone is running a version which does have or is otherwise loading a lot of code into images could you say if it's now fast enough that you don't feel tempted to dump things with loads of stuff preloaded?

I'm asking because I want to know if it's worth recreating & modernising the old infrastructure we had for this, or if it can all be quietly forgotten.

Note I'm talking saving development images, not about delivering applications: obviously you still need to do that.

Thanks

--tim

_______________________________________________
Lisp Hug - the mailing list for LispWorks users
lisp-hug@lispworks.com
http://www.lispworks.com/support/lisp-hug.html

Re: Making images in the modern world

Tim Bradshaw <tfb@tfeb.org> writes:

> Note I'm talking saving development images, not about delivering applications: obviously you still need to do that.

There's built in support for Saved Sessions now that I use instead.

_______________________________________________
Lisp Hug - the mailing list for LispWorks users
lisp-hug@lispworks.com
http://www.lispworks.com/support/lisp-hug.html

Re: Making images in the modern world

Tim Bradshaw wrote on Wed, 4 Jul 2012 20:59:52 +0100 23:59:

| When I last used LW really seriously (between 1999 and 2004 I think) I
| had a hairy infrastructure for making and saving images with code
| preloaded.  I think the purpose of this was to make loading time
| better, especially in the presence of lots of patches and possibly
| large preloaded systems.
|
| Is there any real purpose in doing this now, as opposed to having a
| combination of init files which load things and set logical pathnames
| and (for instance) quicklisp to drag in third-party stuff on demand?
|
| My intuition is that it's not (other than for the special case of
| making a console image) because things are now so fast.

I used to rebuild an image only on when LW patches arrived or my "very
basic" library code changed. It is about once per half a year.

On LW 4.4, loading FASL-files with ASDF and LW defsystem fits my needs quite
well. I have not tried 6.1 but judging by general software engineering
trends, things can be slower there :-)
--
Sincerely,
Dmitriy Ivanov
lisp.ystok.ru

_______________________________________________
Lisp Hug - the mailing list for LispWorks users
lisp-hug@lispworks.com
http://www.lispworks.com/support/lisp-hug.html