I have a situtation where CompareTo is returning strange results when having
dashes in strings. The most concise illustration is:
"10050".CompareTo("1005-") = 1
"10050".CompareTo("1005-0") = -1
I have run alot more tests (I can post a short program if useful), but the
rule seems to be - < 0 at the end of a string but - = 0 in the middle of a
string.

Does anybody know what's going on here?

When comparing strings less than and greater than are very hard to define,
the important thing we need to look at is equality.

--
Sijin Joseph
http://www.indiangeek.net
http://weblogs.asp.net/sjoseph


"Ben Spigle" <Ben (E-Mail Removed)> wrote in message
news:4DDDFC4C-2B91-4A83-9FD4-(E-Mail Removed)...
> I have a situtation where CompareTo is returning strange results when
having
> dashes in strings. The most concise illustration is:
> "10050".CompareTo("1005-") = 1
> "10050".CompareTo("1005-0") = -1
> I have run alot more tests (I can post a short program if useful), but the
> rule seems to be - < 0 at the end of a string but - = 0 in the middle of a
> string.
>
> Does anybody know what's going on here?

Hmm... Maybe you can explain what's so hard about sorting strings. In 25
years in computing I have never before encountered a system that couldn't
sort a list of strings. It's a pretty basic requirement in computing.

I'm trying to sort a list to use in a binary search, where the search key
does not have to be completely specified. Here's how the list I gave it
sorted (quotation marks are just to mark the beginning/end of the key.)
"1005-","10050","1005-0","100500","1005-00","1005-00
","100500 ","100501","1005-01",
"1005-01 ","100501 ","1005-1"
Given a key of "10050" you would hope to get "10050","100500","100500
". With this screwed up sort, however, the algorithm would terminate when
it it saw "1005-0" so neither "100500" and "100500 " would appear in
the result list.
"Sijin Joseph" wrote:

> When comparing strings less than and greater than are very hard to define,
> the important thing we need to look at is equality.
>
> --
> Sijin Joseph
> http://www.indiangeek.net
> http://weblogs.asp.net/sjoseph
>
>
> "Ben Spigle" <Ben (E-Mail Removed)> wrote in message
> news:4DDDFC4C-2B91-4A83-9FD4-(E-Mail Removed)...
> > I have a situtation where CompareTo is returning strange results when
> having
> > dashes in strings. The most concise illustration is:
> > "10050".CompareTo("1005-") = 1
> > "10050".CompareTo("1005-0") = -1
> > I have run alot more tests (I can post a short program if useful), but the
> > rule seems to be - < 0 at the end of a string but - = 0 in the middle of a
> > string.
> >
> > Does anybody know what's going on here?
>
>
>

You're right, actually i was not thinking from that perspective. Anyways i
dug through the BCL code and then through Rotor, i have finally found the
code that String class in .Net uses for string comparison. It is is
\Rotor\sscli\clr\src\classlibnative\nls\comnlsinfo.cpp , go through the code
in this method, it sheds a lot of light on the workings of string
comparison. I haven't had the time to go through the code in detail, but
from what i see there is significant algorithm variance based on case
sensitivity, are you using a case sensitive search?


--
Sijin Joseph
http://www.indiangeek.net
http://weblogs.asp.net/sjoseph


"Ben Spigle" <(E-Mail Removed)> wrote in message
news:BA209C32-FBC5-42D4-B85C-(E-Mail Removed)...
> Hmm... Maybe you can explain what's so hard about sorting strings. In 25
> years in computing I have never before encountered a system that couldn't
> sort a list of strings. It's a pretty basic requirement in computing.
>
> I'm trying to sort a list to use in a binary search, where the search key
> does not have to be completely specified. Here's how the list I gave it
> sorted (quotation marks are just to mark the beginning/end of the key.)
> "1005-","10050","1005-0","100500","1005-00","1005-00
> ","100500 ","100501","1005-01",
> "1005-01 ","100501 ","1005-1"
> Given a key of "10050" you would hope to get "10050","100500","100500
> ". With this screwed up sort, however, the algorithm would terminate
when
> it it saw "1005-0" so neither "100500" and "100500 " would appear
in
> the result list.
> "Sijin Joseph" wrote:
>
> > When comparing strings less than and greater than are very hard to
define,
> > the important thing we need to look at is equality.
> >
> > --
> > Sijin Joseph
> > http://www.indiangeek.net
> > http://weblogs.asp.net/sjoseph
> >
> >
> > "Ben Spigle" <Ben (E-Mail Removed)> wrote in message
> > news:4DDDFC4C-2B91-4A83-9FD4-(E-Mail Removed)...
> > > I have a situtation where CompareTo is returning strange results when
> > having
> > > dashes in strings. The most concise illustration is:
> > > "10050".CompareTo("1005-") = 1
> > > "10050".CompareTo("1005-0") = -1
> > > I have run alot more tests (I can post a short program if useful), but
the
> > > rule seems to be - < 0 at the end of a string but - = 0 in the middle
of a
> > > string.
> > >
> > > Does anybody know what's going on here?
> >
> >
> >