KSChan wrote:
>
>     >>>
>     >> No, and what is worse svn does not fail, it simply ignores any
>     changes
>     >> in svn:externals.
>     >>
>     >> Unless you only ever plan to change docs in a single translation, I
>     >> don't recommend using the svn:externals modules.  However, for
>     those
>     >> people who are exclusively documenting in 1 language, and never
>     touch
>     >> the source code they're documenting it is a great system.
>     >>
>     >> For everyone else, use a sparse checkout.  I put in a small
>     amount of
>     >> effort last week to make phd work, and as long as you aren't
>     adding in
>     >> images, it works fine (phd needs a small change to allow it to find
>     >> images in a sparse checkout, which I have not and will not
>     investigate
>     >> since I don't care about images :).
>     >>
>     >> Greg
>     >>
>     >>
>     >
>     > My experience is different. I'm using TortoiseSVN on Windows XP SP3.
>     >
>     > I recently made a change to doc-en involving both doc-base and en -
>     > see http://svn.php.net/viewvc?view=revision&revision=286553
>     <http://svn.php.net/viewvc?view=revision&revision=286553>
>     >
>
>     Hi,
>
>     Great - so that means users of TortoiseSVN don't have this issue.  The
>     command-line subversion client is the only deficiency if you're using
>     svn:externals.  Good to know.
>
>
> Hi,
>
> I just made a simple test with my repository and native svn-client
>
> Subversion 1.4.4
> native svn-client 1.6.3
>
> repository layout:
>
>     tmp/test1
>     tmp/test2
>     tmp/test3
>
>
> under tmp/test1 , svn:externals
>
>     ext-t2 http://localhost/svn/tmp/test2
>     ext-t3 http://localhost/svn/tmp/test3
>
>
> So, when I make svn co http://localhost/svn/tmp/test1, the whole thing
> becomes
>
>     /test1
>     /test1/ext-t2
>     /test1/ext-t3
>
>
> Then I touched 3 files.
>
>     /test1/t1
>     /test1/ext-t2/t2
>     /test1/ext-t3/t3
>
>
> svn st gives
>
>     X       ext-t2
>     X       ext-t3
>     ?       t1
>
>     Performing status on external item at 'ext-t2'
>     ?       ext-t2/t2
>
>     Performing status on external item at 'ext-t3'
>     ?       ext-t3/t3
>
>
> after svn add and I perform svn ci -m "test" t1 ext-t2/t2 ext-t3/t3 gives:
>
>     Adding         t2
>     Adding         ext-t2/t2
>     Adding         ext-t3/t3
>
>
> Commit with no problem and sharing the same commit log. Thus I think
> svn:externals gives no problem at all (with native svn-client 1.6.x I
> think)

try:

svn commit -m "doesn't work"

in the top-level directory.  All internal svn:externals are ignored.  If
you explicitly provide them, it will work.  Having done this for a few
months, I can say with certainty that 10% or more of the time, one of
the directories with changes will be forgotten in the commit.  We are,
after all, human.

Greg

Reply via email to