Merge Destination on SC 2.0 Beta 1

Hi,

 I'm trying to live with SC 2.0 Beta 1.
It's my pleasure.

 I found an unexpected behave in a merge dialog.
It's about the default at merge destination(undo also).

 My project is the following that.

trunk
branches
tags
AAA(WC)
BBB(WC, current)
CCC(WC)

 I opened a log dialog on BBB in the project view.
And I pushed the Run button.

 I selected two revisions and executed the merge menu.
The merge dialog showed CCC as default at the destination
path area.

 My expectation is BBB. SC 1.2.2 showed it. He is a good
boy.

 On the other case, the current WC is CCC and the log
dialog
is opened on CCC. The merge dialog showed BBB. He is
mischievous.

 On a separate topic, I feel that the blame command is
incomplete on localized encoding files. It is not only for
beta. I expect the behave is like the cat command.

 But I wonder if it is not so easy because the blame output
contains columns besides content. And it's not so important
for me.

 I'm looking forward to use next beta which will support
operations on the repository.
-- 
Kouno

------------------------------------------------------------
---------
To unsubscribe, e-mail: dev-unsubscribesubcommander.tigris.org
For additional commands, e-mail: dev-helpsubcommander.tigris.org

Hi,

Toshihiro Kouno wrote:
> Hi,
> 
>  I'm trying to live with SC 2.0 Beta 1.
> It's my pleasure.
> 
>  I found an unexpected behave in a merge dialog.
> It's about the default at merge destination(undo
also).
> [..]

Yes, that's a bug.


>  On a separate topic, I feel that the blame command is
> incomplete on localized encoding files. It is not only
for
> beta. I expect the behave is like the cat command.

I will take a look at it.

-- 
Martin

Subcommander 1.2.2 - http://subcommander.ti
gris.org
a cross platform Win32/Unix/MacOSX subversion GUI client
& diff/merge tool.

------------------------------------------------------------
---------
To unsubscribe, e-mail: dev-unsubscribesubcommander.tigris.org
For additional commands, e-mail: dev-helpsubcommander.tigris.org

Hi,

Toshihiro Kouno wrote:

> On a separate topic, I feel that the blame command is
> incomplete on localized encoding files. It is not only
for
> beta. I expect the behave is like the cat command.

I'm getting crazy over this stuff 

For me, cat is broken again. Looks like they have changed
the
conversion stuff in svn_client_cat. Some time ago cat
delivered
the file in "locale" encoding and i had to convert
it back to
utf8 before letting qt convert it to ucs2/utf16 for
display.

Now, svn_client_cat returns the file in utf-8. What it
should
have done from the beginning. After removing the extra
conversion
cat works as expected.


>  But I wonder if it is not so easy because the blame
output
> contains columns besides content. And it's not so
important
> for me.

Blame is another case. Subcommander gets everything
properly
encoded in utf-8 from subversion and it gets converted to
ucs2
for display, everything perfect.

The strange thing is now that from the following 4 lines
the
first 3 are displayed correctly and the last not. If the
line
start with non japanese characters it is displayed fine, if
it
starts with a japanese character it just shows boxes.

öäöäöäöäöäöäöüüüßßßßßßさくらさく�
��さくらさくらさくらさくらさくらさくらßß
ßßßßßßßßßß
öさくらさくらさくらさくらさくらさくらさ�
��らさくら
aさくらさくらさくらさくらさくらさくらさ�
�らさくら
さくらさくらさくらさくらさくらさくらさく
らさくら

I think that is a bug in the (ancient) qt version I'm using
on
Windows. This will hopefully go away when Subcommander
moves
from qt3 to qt4 (after 2.0.0 is done...).

-- 
Martin

Subcommander 1.2.2 - http://subcommander.ti
gris.org
a cross platform Win32/Unix/MacOSX subversion GUI client
& diff/merge tool.


------------------------------------------------------------
---------
To unsubscribe, e-mail: dev-unsubscribesubcommander.tigris.org
For additional commands, e-mail: dev-helpsubcommander.tigris.org

Hi,

> For me, cat is broken again.

 I feel that cat is not broken. I can get the expected view
for Japanese encoding files on both SC(2.0/1.2.2).

> Some time ago cat delivered the file in
"locale" encoding

 It seems to continue now. I think that the "svn"
command
line tool has no conversion at cat in SVN(1.4.3) source.

> Now, svn_client_cat returns the file in utf-8.

 I should think that this file was utf-8 in the repository
and before adding to the repository.

> Blame is another case. Subcommander gets everything
properly
> encoded in utf-8 from subversion and it gets converted
to ucs2
> for display, everything perfect.

 I feel that SC is getting a content by direct encoding.
Current SC blame output seems be like past SC cat output.

> The strange thing is now that from the following 4
lines the
> first 3 are displayed correctly and the last not. If
the line
> start with non japanese characters it is displayed
fine, if it
> starts with a japanese character it just shows boxes.

 Hmm... I cannot reproduce this strange state.

 I show my trial. My Subversion is 1.4.3.

(1) consider a test file

 This test file is utf-8 and contains two region non-ASCII
characters. Japanese usual encodings cannot fully represent
this file content.

 I prepared three test files and add to my repository.

(a) original utf-8(full)
(b) utf-16(full)
(c) Japanese Windows encoding(replaced some characters)

(2) cat and blame

 'cat'(SC) show incorrect views for (a) and (b). (c) is no
problem. The "svn" tool is similar.

 'blame'(SC) says that this file is binary for (a) and (b).
(c) is an incorrect view(like old 'cat'). The
"svn" tool is
similar, but it shows a correct view for (c).

 (a) and (b) have mime-type
property("octet-stream").

(3) without mime-type

 I removed mime-type. But the result is same thing.

 I tried "svn add --no-auto-props". It said
"A (bin)
filename"(usually "A filename"). And the
result is same
thing.

 This is my trial. I could not reproduce the strange state.
I did only on Japanese Windows and SC 2.0 Beta.
-- 
Kouno

------------------------------------------------------------
---------
To unsubscribe, e-mail: dev-unsubscribesubcommander.tigris.org
For additional commands, e-mail: dev-helpsubcommander.tigris.org