Re: [Gcl-devel] Compiling GCL

[Faré]

Ping?

One function that fails to return the same thing when compiled vs
interpreted is REDUCE/STRCAT from uiop/utility.lisp.

Another function that behaves in a weird way is configure-asdf from
test/script-support.lisp, whereby the form
(acall :pathname-directory-pathname (nth-value 2 (acall :locate-system
:test-asdf)))
returns NIL when (load-asdf) is run, vs the proper answer (the
pathname of .../asdf/test/) when (configure-asdf) is run directly.
Possibly, the fact that the functions are being (re)defined while
(load-asdf) is run is confusing GCL. Or maybe there's a problem with
multiple-values in some operation mode.

Can you look into it? There are other failures, but these are show stoppers.

—♯ƒ • François-René ÐVB Rideau •Reflection&Cybernethics• http://fare.tunes.org
The War on Terrorism is missing the point: what we need is a War on War!
        — Kennita Watson


On Sat, Nov 9, 2013 at 3:58 PM, Faré <address@hidden> wrote:
> On Thu, Nov 7, 2013 at 9:15 AM, Camm Maguire <address@hidden> wrote:
>>> Logical pathnames are both tricky and ultimately useless. They are a
>>> legacy thing that was never developed (premature standards cause
>>> premature code that can't evolve) and is now replaced by better
>>> things. I wouldn't make a priority of getting them perfect.
>>
>> Thanks so much for saying this!  You seem to make heavy use of it for
>> some reason in your tests, upon seeing which I thrust my hands
>> heavenward and exclaim 'no, not that.... anything but that...!' :-)
>>
> Yeah, well, as you can see, I procrastinated on more important stuff
> by obsessing about getting things working right with logical pathnames,
> for the benefit of the handful (max) of misguided cranky semi-old timers
> who use them and report bugs. janderson, p-cos — who else is crazy enough
> to use "logical" pathnames?
>
> Because I was semi-regularly getting reports that I was breaking support
> for logical-pathnames, I ended up including them in my tests.
> They are also in half the tests in the janderson-contributed
> asdf-pathname-test.script.
>
> Back to the failures.
> 1- Even simple uses of STRCAT seem to bork:
> (uiop:strcat "abc" "def") should return "abcdef" but instead fails
> Even (uiop:strcat "abc") fails mysteriously.
> In compiling etypecase, the type string seems badly supported.
>
> 2- there are plenty of condition-related issues.
> I haven't been able to isolate then, but shouldn't be too hard
> to reproduce with the instructions given by the test suite on failure.
> It seems to me that conditions are not compiled properly,
> perhaps due to all the eval-when business.
>
> 3- Last but not least, there are pathname issues that are not related
> to logical pathnames.
>  I admit I am not too familiar with that part of the test suite.
>
> I'll try again when I have a decent machine and connection (I'm
> currently on the move).
>
> —♯ƒ • François-René ÐVB Rideau •Reflection&Cybernethics• http://fare.tunes.org
> Many people escape old fairy tales (creationism, homeopathy, monarchy)
> only to fall into new ones (democracy, socialism, global warmism)...