Ada Structure Library 1.4 release

Ada Structure Library 1.4 release

[Corey Minyard]

I have just put a new version of AdaSL on SourceForge 
(http://adasl.sf.net).  This add the following:

 * A reference counting pointer
 * A rework of the string tokenizer to make it more usable.
 * A calendar package

The biggie here is the calendar package.  It does pretty much anything 
you want with a Gregorian calendar, including leap seconds.  If you do 
fancy processing across timezones or back in time, this is the package 
for you.  I'd appreciate any commentary on this, like ease of use, 
understandability, etc.

Also, it generates a timezone file from the zone info files supplied 
with glibc.  The trouble is that the full generated file is huge (8800 
lines, about 1/2 meg).  It contains all the timezones you could possibly 
imagine back to when timezones started.  There is a much smaller 
simplified version that only contains the current timezone data (no 
historical information).  I'm curious what people think I should do with 
the huge file.  I could put the information in files and read it in on 
demand, but then the system has to have files go along with it.  I could 
break it up to continent chunks, but that doesn't seem to gain much and 
complicates things.  Just curious if anyone has any ideas.

-Corey

Re: Ada Structure Library 1.4 release

[Nick Roberts]

"Corey Minyard" <minyard@acm.org> wrote in message
news:3BF1887F.5020307@acm.org...
> ..
> Also, it generates a timezone file from the zone info files supplied
> with glibc.  The trouble is that the full generated file is huge (8800
> lines, about 1/2 meg).  It contains all the timezones you could possibly
> imagine back to when timezones started.  There is a much smaller
> simplified version that only contains the current timezone data (no
> historical information).  I'm curious what people think I should do with
> the huge file.  I could put the information in files and read it in on
> demand, but then the system has to have files go along with it.  I could
> break it up to continent chunks, but that doesn't seem to gain much and
> complicates things.  Just curious if anyone has any ideas.

On a typical modern workstation (or PC), a 0.5 MB file is tiddly. Absolutely
no problem. Is it in binary format? If not, perhaps it could be made more
compact anyway.

On any machine for which such a file is too big, it's very unlikely that
full historical timezone computations would be required. More likely, only a
very simple time model would be required.

Problem solved!

--
Best wishes,
Nick Roberts


PS: The concept of a 'continent chunk' would give any gastro-urinary tract
specialist a nightmare, I suspect! ;-)

Re: Ada Structure Library 1.4 release

[Corey Minyard]

Thanks for the input.  That's kind of what I thought, too, but Ada plays 
in a lot of different places, and I wanted some input.

-Corey

Nick Roberts wrote:

> "Corey Minyard" <minyard@acm.org> wrote in message
> news:3BF1887F.5020307@acm.org...
> 
>>..
>>Also, it generates a timezone file from the zone info files supplied
>>with glibc.  The trouble is that the full generated file is huge (8800
>>lines, about 1/2 meg).  It contains all the timezones you could possibly
>>imagine back to when timezones started.  There is a much smaller
>>simplified version that only contains the current timezone data (no
>>historical information).  I'm curious what people think I should do with
>>the huge file.  I could put the information in files and read it in on
>>demand, but then the system has to have files go along with it.  I could
>>break it up to continent chunks, but that doesn't seem to gain much and
>>complicates things.  Just curious if anyone has any ideas.
>>
> 
> On a typical modern workstation (or PC), a 0.5 MB file is tiddly. Absolutely
> no problem. Is it in binary format? If not, perhaps it could be made more
> compact anyway.
> 
> On any machine for which such a file is too big, it's very unlikely that
> full historical timezone computations would be required. More likely, only a
> very simple time model would be required.
> 
> Problem solved!
> 
> --
> Best wishes,
> Nick Roberts
> 
> 
> PS: The concept of a 'continent chunk' would give any gastro-urinary tract
> specialist a nightmare, I suspect! ;-)