Re: [Chicken-hackers] Outstanding patches

[Peter Bex]

On Tue, Nov 13, 2012 at 07:50:44PM +0100, Felix wrote:
> > "always use temporary filenames in csc"
> > http://lists.nongnu.org/archive/html/chicken-hackers/2012-11/msg00017.html
> > 
> > This discussion is still ongoing, but I wanted to mention for completeness.
> > 
> 
> I will provide a patch that checks for the files given on the command line.
> This should be enough for the time being.

Agreed.

> > "install-extension fix for when the second arg contains pairs (SOURCE DEST)"
> > http://lists.nongnu.org/archive/html/chicken-hackers/2012-10/msg00049.html
> > 
> > I think this can probably just be applied as-is.  I remember Felix looked
> > at this at T-DOSE but he said he needed to have a closer look, IIRC.
> 
> Yes, this seems to be ok (I haven't looked).

I've now pushed this.

> > "Add .gitignore"
> > http://lists.nongnu.org/archive/html/chicken-hackers/2012-10/msg00048.html
> > 
> > I'm unsure.  If it bothers Mario enough, maybe it should just be applied.

Mario mentioned he didn't care that much.  I think he put it on the wiki
for people who want this (to use in ~/.gitignore).

> > "disallow compiling from both stdin & named files"
> > http://lists.nongnu.org/archive/html/chicken-hackers/2012-10/msg00033.html
> > 
> > Unsure.  Maybe this will get resolved automatically when the temporary
> > files thread gets resolved.
> 
> I think so, too.

OK, I'll leave it at that then.

> > "'deep-stack' option, declaration and build-mode"
> > http://lists.nongnu.org/archive/html/chicken-hackers/2012-08/msg00002.html
> > 
> > I don't like this much as it doesn't improve the situation "by default".
> > Asking newbies to recompile Chicken (this patch seems mostly aimed at
> > newbies and Scheme purists) is overly burdensome.  Maybe we should
> > just forget about it.
> 
> This feature is not for newbies, it's for purist who know what they
> ask for. It has no impact on the default build, so I see no reason
> not to add it.

I'm unsure of the usefulness of this option if it's disabled by default.
The original context was someone who asked on IRC why his SICP exercise
program was segfaulting, so that's definitely a newbie.

My reasoning is that it makes more sense to have it enabled by default
with a switch to disable it for performance reasons (like most other
performance optimizations; they're also disabled by default!) than to
have it the other way around.  The biggest problem with that is of
course the fact that this has an impact on the runtime library that you
can't easily get rid of unless you recompile.

If anyone else thinks it's useful enough to add even if disabled by
default, feel free to pipe up or push it.  I don't care much for yet
another compiler option and build-time switch.  Most people will install
chicken via their (binary) package manager, so a build time switch won't
be very useful unless they really are such crazy purists that they are
willing to build a new Chicken with this option in the runtime library.
I think such rabid purists are unlikely to be using Chicken anyway.

*makes an effort to silence his inner rabid purist*

If we later decide we need it for whatever reason, it'll be in the mail
archives.

> > Windows is the main reason that we haven't applied it as-is.  Maybe one
> > of our Windows users can have another look at how to fix this?
> > The patch works for *nix just fine.  I'm not 100% sure about platforms
> > like Haiku.  Maybe we should just use poll() on platforms that support it?
> 
> It won't work on Windows, and the patch makes assumptions about the
> layout of the fdset that do not apply to Windows (the fdset is just
> a wrapper for some Win32 data structure, IIRC). For the same reason
> it may fail on other platforms.

Goddamn that shitty excuse for an OS.  I've went ahead and posted a
patch that takes care of this by using poll() instead of select() where
available.  I hope this is deemed acceptable.

I think we've now taken care of all these patches!

Cheers,
Peter
-- 
http://sjamaan.ath.cx
--
"The process of preparing programs for a digital computer
 is especially attractive, not only because it can be economically
 and scientifically rewarding, but also because it can be an aesthetic
 experience much like composing poetry or music."
                                                        -- Donald Knuth