On Mar 16, 2005, at 8:30 AM, Ben Collins-Sussman wrote:
>
> On Mar 16, 2005, at 8:21 AM, Brian W. Fitzpatrick wrote:
>
>> Is there a reason that it does this? I've attached a patch that
>> removes it, but wanted to run it past someone else (sussman?
>> lundblad?) to see if there was a deep dark reason that we always
>> wanted an empty line at the beginning of lslock's output. Patch to
>> fix attached (I'll commit it if no one objects).
>>
>
> It looks fine to me.
>
> But: let me change the subject.
>
> Mike Pilato pointed out a problem with all three commands that display
> lock details:
>
> svnadmin lslocks
> svnlook lock
> svn info wcpath-or-URL
>
> Each of those commands shows the lock-comment field last, because it's
> the only one that's potentially a multi-line value.
>
> The problem is that svn's output is always supposed to be machine
> parseable. Up till now, the only time we've had to deal with
> unpredictable multi-line output is with 'svn log', and the problem was
> solved by printing a header saying "N lines", so a parser knows how
> many lines of commit-message to slurp up.
>
> So Mike's pointed out that we should probably be doing the same thing
> for those three commands -- right? Heck, maybe even a separator line
> between objects would help too, just like we do for 'svn log'.
>
> Do others concur?
+1
-Fitz
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Wed Mar 16 15:56:29 2005