8:22 a.m.
Hi,
simple script:
<hansl>
bundle foo1
string foo1.str1 = ""
string foo1.str2 = "~"
bwrite(foo1, "foo1")
foo2 = bread("foo1")
eval foo1
eval foo2
</hansl>
In foo1 we have empty string str1 but foo2 doesn't contain it. I checked
the XML file in wchich bwrite stores the foo1 bundle: there is no
definition of str2 anymore.
Is this intentional behaviour or bug?
Regards,
Marcin
--
Marcin Błażejowski
8:44 a.m.
Am 30.08.2017 um 10:22 schrieb Marcin Błażejowski:
In foo1 we have empty string str1 but foo2 doesn't contain it.
I'd say it's a bug.
the XML file in wchich bwrite stores the foo1 bundle: there is no
definition of str2 anymore.
You mean str1? Because str2 survives OK for me
(
? eval foo2
bundle foo2:
str2 = ~
)
BTW, there may be an additional small bug. The doc in chapter 14.2 says
about a string declaration: "This can be the keyword null, to produce an
empty string." However, this doesn't work in the bundle context, try this:
bundle b
string b.check = null # fails
However, perhaps we don't want to have more of this "null" business, and
perhaps it should rather be deprecated for strings, given that empty
strings are already a good way of defining a non-existent string type?
cheers,
sven
8:14 p.m.
On Wed, 30 Aug 2017, Sven Schreiber wrote:
BTW, there may be an additional small bug. The doc in chapter 14.2
says
about a string declaration: "This can be the keyword null, to produce an
empty string." However, this doesn't work in the bundle context, try this:
bundle b
string b.check = null # fails
However, perhaps we don't want to have more of this "null" business, and
perhaps it should rather be deprecated for strings, given that empty
strings are already a good way of defining a non-existent string type?
Agreed. I've removed that snippet of text from the Guide. "null"
will still work (for plain strings) but there's no call to recommend
it over empty quotes ("").
Allin
8:08 p.m.
On Wed, 30 Aug 2017, Marcin Błażejowski wrote:
Hi,
simple script:
<hansl>
bundle foo1
string foo1.str1 = ""
string foo1.str2 = "~"
bwrite(foo1, "foo1")
foo2 = bread("foo1")
eval foo1
eval foo2
</hansl>
In foo1 we have empty string str1 but foo2 doesn't contain it. I checked
the XML file in wchich bwrite stores the foo1 bundle: there is no
definition of str2 anymore.
Is this intentional behaviour or bug?
Well, if you were to watch stderr you'd see:
bundle -> XML: skipping empty string str1
However, maybe we shouldn't be doing that.
Allin
4:28 a.m.
W dniu 30.08.2017 o 22:08, Allin Cottrell pisze:
Well, if you were to watch stderr you'd see:
bundle -> XML: skipping empty string str1
Wright.
But in my case it looks like a bug. I use bwrite/bread to communicate
between regular gretl instance and MPI world, so if I loose already
defined item in bundle my programs crashes. I found workaround for it
but only for my case.
Regards,
Marcin
--
Marcin Błażejowski
3:33 p.m.
On Fri, 1 Sep 2017, Marcin Błażejowski wrote:
W dniu 30.08.2017 o 22:08, Allin Cottrell pisze:
> Well, if you were to watch stderr you'd see:
>
> bundle -> XML: skipping empty string str1
Wright.
But in my case it looks like a bug. I use bwrite/bread to communicate
between regular gretl instance and MPI world, so if I loose already
defined item in bundle my programs crashes. I found workaround for it
but only for my case.
In current git empty strings in a bundle will be preserved via
bwrite/bread.
Allin
2673
days inactive
2675
days old
5 comments
3 participants
participants (3)
-
Allin Cottrell -
Marcin Błażejowski -
Sven Schreiber