View Issue Details

IDProjectCategoryView StatusLast Update
0000943MetaPostfeature requestpublic2015-08-04 18:02
Reportertoby Assigned To 
Status newResolutionopen 
Product Version1.890 
Summary0000943: in, eps and infinity constants could be better defined for new number systems
DescriptionCurrently has these three definitions...

eps := .00049; % this is a pretty small positive number
epsilon := 1/256/256; % but this is the smallest
infinity := 4095.99998; % and this is the largest

it would be nice if the decimal constants could be replaced by expressions that could be evaluated at full precision by the new number systems. Like this:

eps := 1/2048; % this is a pretty small positive number
epsilon := 1/256/256; % but this is the smallest
infinity := 64*64-epsilon; % and this is the largest

This would make no difference when using the default scaled numbers, but would make the constants more consistent when using higher precision number systems.

In particular it would make "infinity+epsilon=4096" true which ever number system is used.
TagsNo tags attached.



2015-08-04 17:16

reporter   ~0001393

While we are about it, it would also be nice to replace the units of measure definitions with equivalent relations that are more accurate with the higher precision numeric engines. For example replace the existing lines in ""

mm=2.83464; pt=0.99626; dd=1.06601; bp:=1;
cm=28.34645; pc=11.95517; cc=12.79213; in:=72;

with these equivalent lines, which give full precision regardless of engine

72 = 1 in; 800 = 803 pt; 360 = 127 mm; 3600 = 127 cm;
1 pc = 12 pt; 1 cc = 12 dd; 1157 dd = 1238 pt; bp = 1;

luigi scarso

2015-08-04 18:02

manager   ~0001394

Hm, I tend to consider kind of frozen --- I think it could be still in use for TAOCP, for example. Better to think to a instead, or something similar.

Issue History

Date Modified Username Field Change
2015-07-27 20:24 toby New Issue
2015-08-04 17:16 toby Note Added: 0001393
2015-08-04 18:02 luigi scarso Note Added: 0001394