On 11.12.2015 06:19, Andy Bradford said:

>> when called  as a  process [1].  I don't  know if  this can  be solved
>> inside fossil; a workaround is to use a modified plink, e.g. that from
>> TortoiseSVN.
>
> You can configure Fossil to use the modified plink. Use:
>
> fossil clone --ssh-command /path/to/modified/plink.exe -T -e none ...

That is exactly what I've done (and more, i.e. "fossil settings --global
ssh-command PATH").

>
>> Still there seems to be another  problem with fossil: it does not pass
>> the password  to plink  when it was  given on the  command line  as in
>> user:pass@host:port.
> This is  because Fossil does not  interact with SSH, the  end user does.
> Fossil  forks an  SSH  command, and  you, the  user,  interact with  any
> prompts  the  SSH  process  issues. When  you  have  completed  entering
> password information into  SSH, Fossil now has a set  of encrypted pipes
> to read/write to via stdin/stdout.

I agree. Still the documentation (e.g. fossil clone) mentions this
possibility for ssh URL's ([userid[:password]@]host), so in my opinion
either fossil passes the password further to plink (it cannot do this on
Linux to ssh, since that one has no password argument), or it removes
this altogether from documentation.

Would it be an idea to detect the case Windows and no Pageant (or maybe
add some new "-p" fossil argument) and implement password prompt inside
fossil? Just contemplating...

Side note: as for the security risk, I agree in principle, but since the
user has already decided to type in his password on fossil's command
line, the evil is there and passing it to plink makes it no worse.

Daniel
_______________________________________________
fossil-users mailing list
fossil-users@lists.fossil-scm.org
http://lists.fossil-scm.org:8080/cgi-bin/mailman/listinfo/fossil-users

Reply via email to