[Gcl-devel] Re: Comment About Room

[Camm Maguire]

Greetings!

Robert Boyer <address@hidden> writes:

> Hi Camm,
> 
> If you would like us to fix your format issues, then change the protection on
> your "twc+" directory and point us to the function that prints this
> information.
> 

OK, in /u/camm/gcl-2.6.6twc+/lsp/gcl_autoload.lsp, the function (room)
(and (heaprep)).

I appreciate your help on this, as I'm particularly bad at user
interface type questions.

> With respect to the report, the output after the line 
> 
>     262144 maximum pages
> 
> is good information but doesn't (?) change after GCL is started.  Therefore,
> it seems like this information need only be printed upon "special" request,
> e.g., on (room t) but not (room nil) or just (room).  Here are some phrasing
> changes that we suggest.
> 

Or perhaps as a separate function, which could be documented.  I just
thought that questions of the sort 'why can't my heap grow further',
and 'why can't I recurse deeper" come up frequently enough that this
information should be brought to the attention of the user somewhere,
and everyone is used to (room).

> WS: 32-bit words per structure of the type indicated

type -> types (likewise below), and '32' being variable, then I agree.

> UP: currently allocated pages of the type indicated
> MP: maximum current number of pages permitted to be allocated for the type 
> indicated

Suggest current maximum

> FI: fraction of cells in use on allocated pages of the type indicated 
> immediately 
>     after the last garbage collection (si::gbc t)
> 

Technically, this is the fraction of cells on the allocated pages,
with the remainder being free objects.  Doing a fresh gc will let you
know what you are really holding onto, but the number has some meaning
elsewhen.  The number of pages allocated could have grown since the
last gc as well, so I'm not sure about the wording above.

> It would also be nice if the output from this report was never longer than
> about 70 or 75 characters long per line.
> 

How about wrapping the types, or do you prefer truncation?

Take care,

> Bob & Warren
> ++++++++++++
> 
> ACL2>(room)
> 
> WS      UP/MP     FI%      GC TYPES
> 
>  2   40000/40000  49.2%       4CONS FIXNUM SHORT-FLOAT CHARACTER READTABLE 
> SPICE
> 10     252/351  99.8%       1 SYMBOL
> 14       1/2    39.7%         PACKAGE
>  8    2000/2000  51.8%        ARRAY HASH-TABLE VECTOR BIT-VECTOR STREAM 
> PATHNAME CCLOSURE CLOSURE
>  4   10000/10000   1.0%       STRUCTURE BIGNUM RATIO LONG-FLOAT COMPLEX CFUN
>  6     388/1032  99.8%       2SFUN STRING RANDOM-STATE GFUN VFUN AFUN CFDATA
> 
>       2000/2000               contiguous (290 blocks)
>            4550               hole
>            60000  88.1%       1relocatable
> 
>      52641 pages for cells
>     119191 total pages
>     106831 pages available
>      36122 pages in heap but not gc'd + pages needed for gc marking
>     262144 maximum pages
> 
> 
> Key:
> 
> WS: words per struct
> UP: allocated pages
> MP: maximum pages
> FI: fraction of cells in use on allocated pages
> GC: number of gc triggers allocating this type
> 
> word size:            32 bits
> page size:            4096 bytes
> heap start:           0x0
> heap max :            0x40000000
> shared library start: 0x40000000
> cstack start:         0xc0000000
> cstack mark offset:   212 bytes
> cstack direction:     downward
> cstack alignment:     16 bytes
> cstack max:           1073741824 bytes
> immfix start:         0xc0000000
> immfix size:          536870912 fixnums
> 
> ACL2>
> 
> 
> 

-- 
Camm Maguire                                            address@hidden
==========================================================================
"The earth is but one country, and mankind its citizens."  --  Baha'u'llah