On Thu, 16 Jan 2014, Hélio Guilherme wrote:
On Thu, Jan 16, 2014 at 11:12 PM, Riccardo (Jack) Lucchetti <
r.lucchetti(a)univpm.it> wrote:
> On Thu, 16 Jan 2014, Allin Cottrell wrote:
>
> Here's the "RFC" I promised in response to Sven's posting at
>>
http://lists.wfu.edu/pipermail/gretl-devel/2014-January/004867.html
>>
>
> [...]
>
> This little story (which admits several variations on the same theme) is
> perhaps a bit stretched here and there; and sure, there is some
> carelessness involved, but losing all the data for a project out of one or
> two silly moves is unacceptable. IMO, there should be some mechanism by
> which the potential for confuson is kept at a minimum. The openoffice
> format, for example, is a zip file containing several xml files. How about
> something similar? "Foo.bdt" may be a zip file containing pre-defined xml
> files: "data.gdt" (perhaps containing some additional metadata) and
> "data.bin" (the binary blob); we could use a "light" compression
level so
> to preserve decent speed when reading and writing.
>
I totally agree with this solution, a zipped file containing metadata and
data .gdt Xor .bin. Gretl could use the same file gdt extension for the
package, even could include, for example, info about the binary content
like MD5 checksum (which is a time consuming operation but you may consider
if it is needed).
I also agree, this is an idea worth pursuing. The main problem is
that the two bits of a binary data file could come unstuck, and
Jack's proposal solves that. It also justifies having a different
filename extension: this really would be a different sort of data
file. I'll try some experimentation.
Allin