Re: st: Mata buffered read 64-bit integer (8 byte/int64), how?

[Sergiy Radyakin <serjradyakin@gmail.com>]

Yes, Stas, this is very close to what I want, though -quadsum()- is
using quad precision reals (float=4, double=8, quad=16 bytes), while
for me 64-bit integers (8 bytes) are sufficient.

And importantly, in Stata 9.

http://en.wikipedia.org/wiki/Quadruple-precision_floating-point_format
http://en.wikipedia.org/wiki/Int64


Sergiy.

On Thu, Jun 13, 2013 at 5:20 PM, Stas Kolenikov <skolenik@gmail.com> wrote:
> I guess at least a part of Sergiy's question is, "Can the mere mortals
> like himself use the quad type that is obviously out there in Mata, as
> all of the -quadsum()- and the like functions provide at least some
> interface to this type?"
>
>
> -- Stas Kolenikov, PhD, PStat (SSC)
> -- Senior Survey Statistician, Abt SRBI
> -- Opinions stated in this email are mine only, and do not reflect the
> position of my employer
> -- http://stas.kolenikov.name
>
>
>
> On Thu, Jun 13, 2013 at 3:55 PM, Sergiy Radyakin <serjradyakin@gmail.com> wrote:
>> Thank you, Stas,
>> this is what I am doing now. Storage in a struct (or in my case simply
>> in two real variables) is not a problem. The problem starts when you
>> need one (LARGE) value, then x=x1*2^32+x2 becomes already lossy, since
>> (as the manual says) : "Doubles can store up to 9,007,199,254,740,991
>> exactly". Not that I would expect anyone to have files that big :)
>> but formally I want to be on the safe side and I didn't expect a
>> problem having 64-bit integers in (at least) 64-bit Stata.
>>
>> Sergiy
>>
>> On Thu, Jun 13, 2013 at 4:48 PM, Stas Kolenikov <skolenik@gmail.com> wrote:
>>> You can probably define the -struct-ure that could store such strings,
>>> and convert them into numbers for arithmetic operations. Bill Gould
>>> may have had an article on this in Stata Journal -- the generic use of
>>> structures is described in
>>> http://www.stata-journal.com/sjpdf.html?articlenum=pr0035. Coding
>>> matrix multiplication with these elements will quickly become an
>>> overkill :-\.
>>>
>>> -- Stas Kolenikov, PhD, PStat (SSC)
>>> -- Senior Survey Statistician, Abt SRBI
>>> -- Opinions stated in this email are mine only, and do not reflect the
>>> position of my employer
>>> -- http://stas.kolenikov.name
>>>
>>>
>>>
>>> On Thu, Jun 13, 2013 at 3:12 PM, Sergiy Radyakin <serjradyakin@gmail.com> wrote:
>>>> I read a 32-bit integer value from a buffered file with d=fbufget(C,fh,"%4bu").
>>>>
>>>> How do I read a 64-bit integer from the same file when Mata does not
>>>> support format "%8bu"? And what can be the type of the LHS (type of
>>>> variable d above) to accommodate the result of the read without the
>>>> loss of information (double is not good, not sure about quad
>>>> precision).
>>>>
>>>> I can read it as 8-bytes string, or 8 individual bytes, but I also
>>>> need to work with the result as a value in Mata. So looking for a
>>>> convenient solution.
>>>>
>>>> Thank you,
>>>>      Sergiy
>>>> *
>>>> *   For searches and help try:
>>>> *   http://www.stata.com/help.cgi?search
>>>> *   http://www.stata.com/support/faqs/resources/statalist-faq/
>>>> *   http://www.ats.ucla.edu/stat/stata/
>>> *
>>> *   For searches and help try:
>>> *   http://www.stata.com/help.cgi?search
>>> *   http://www.stata.com/support/faqs/resources/statalist-faq/
>>> *   http://www.ats.ucla.edu/stat/stata/
>> *
>> *   For searches and help try:
>> *   http://www.stata.com/help.cgi?search
>> *   http://www.stata.com/support/faqs/resources/statalist-faq/
>> *   http://www.ats.ucla.edu/stat/stata/
> *
> *   For searches and help try:
> *   http://www.stata.com/help.cgi?search
> *   http://www.stata.com/support/faqs/resources/statalist-faq/
> *   http://www.ats.ucla.edu/stat/stata/
*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/faqs/resources/statalist-faq/
*   http://www.ats.ucla.edu/stat/stata/