Character sets misrepresented

Maxim · Post by **Maxim** » Thu Jul 19, 2007 11:47 am

We have some non-standard characters in code (i.e. "ç"), and we've noticed they are represented by two characters when we diff by using the Vault client.

Here's an example:

Now, we are getting the right characters when getting the latest version of the files. Is there any potential problem with this?

Beth · Post by **Beth** » Thu Jul 19, 2007 1:24 pm

Are they only showing that way in the diff/merge tool or do you get two different results when you are using what you normally do for editing?

Maxim · Post by **Maxim** » Thu Jul 19, 2007 2:15 pm

I'm getting the right characters (using Visual Studio 2005). No problems so far.

I'm just thinking ahead, wanting to avoid future problems. So might this be just a matter of character representation in the Vault client?

Maxim · Post by **Maxim** » Fri Jul 20, 2007 7:03 am

Beth wrote:Are they only showing that way in the diff/merge tool or do you get two different results when you are using what you normally do for editing?

The Vault client's diff screen is showing them that way (image above).

Beth · Post by **Beth** » Fri Jul 20, 2007 8:16 am

What version of Vault are you using?

I think what may take care of your issue is the new DiffMerge tool we have available separately: http://www.sourcegear.com/diffmerge/downloads.html.

Then in your Tools - Options - DiffMerge, you only need to change the path to point at the new tool.

Maxim · Post by **Maxim** » Fri Jul 20, 2007 8:29 am

Beth wrote:What version of Vault are you using?

3.5.1 (4786).

Beth wrote:I think what may take care of your issue is the new DiffMerge tool we have available separately: http://www.sourcegear.com/diffmerge/downloads.html.

Then in your Tools - Options - DiffMerge, you only need to change the path to point at the new tool.

I've tried the new DiffMerge tool and it shows the same two characters for the "ç" character.

However, we are not having any issues really. We only have concerns this could be a Vault problem handling different character sets, even though we are getting the right characters when we get the files from the Vault repositories.

My question is, since we are getting back the same characters we are storing but in the Vault programs they are showed different, could this be a character representation bug in the Vault programs? Are we safe?

jeffhostetler · Post by **jeffhostetler** » Fri Jul 20, 2007 9:03 am

it looks like DiffMerge loaded the file assuming a latin-n character
encoding when it appears that the file is utf-8. the 2 characters
displayed in the screenshot are a UTF8 multi-byte sequence for an
accented character.

using the new version of DiffMerge, check the Ruleset (via the
Options dialog) used for files with that suffix and change it to
UTF8 (it's probably currently set to ISO-Latin-1). you may also
want to change the default ruleset as well.

then exit and restart DiffMerge and see if the file appears correctly.

to answer your other questions, vault does not need to look at the
character encoding to send/receive/store files. it does not change
files when sending/receiving/storing, so your files are safe.

this problem is limited to DiffMerge and how it interpreted the files
as it read them for display. in this case, the file was in a different
encoding which caused the conversion to unicode for display to be
wrong.

i hope i've answered all your questions and sorry for the confusion.
jeff

Maxim · Post by **Maxim** » Fri Jul 20, 2007 9:51 am

Thanks. I added the file suffix, set its enconding to UTF-8, and it showed correctly. It was a presentation issue.

Btw, it would be great if you could add the XML suffix by default -widely used nowadays.

Thanks again!

jeffhostetler · Post by **jeffhostetler** » Fri Jul 20, 2007 10:02 am

thanks for the suggestion.

i'll log an a feature request to add an xml ruleset to the initial
set of rulesets.

j

SourceGear Support

Character sets misrepresented

Character sets misrepresented

this is a UTF8 issue

thanks for the suggestion