SVN Problems

[James E Keenan]

Andrew Whitworth wrote:
> Last night while trying to merge the kill_parrot_cont branch, I got a
> strange SVN error. 

Today, cotto got the same thing while trying to
> merge pluggable_runcore. Here's what we are getting:
> 
> svn: REPORT of '/parrot/!svn/vcc/default': Could not read chunk size:
> Secure connection truncated (https://svn.parrot.org)


I came on line toward the end of the period Saturday evening during 
which you were experiencing problems, and I also experienced problems, 
albeit with different error messages:

svn: Could not get content-type from response

svn: Server sent unexpected return value (304 Not Modified) in response 
to GET request for '/parrot/!svn/ver/41024/branch​ 
es/kill_parrot_cont/src/hll.c' [This was observed with other source 
files mentioned, as well]

svn: Error reading spooled REPORT request response

I, too, came up with little that was clear-cut upon internet search.  It 
was not a problem with branching/merging per se, because I was able to 
create a branch, make a single trivial modification, then merge branch 
into trunk (in sandbox) and commit to trunk.  I do not think it was a 
problem with any particular file in the merge; we examined svn 
properties, etc., for several of them.

Various hypotheses were hypothesized:

1.  Cert-related problems.  When a branch is created and then your cert 
expires, do you have problems subsequently merging?

2.  Size of merge.  Well, we've had bigger merges re numbers of files, 
but the number of lines involved was in the 5-figure range.

3.  Temporary Apache problems?

4.  Number of files in conflict?  There was a brief window wherein I was 
able to do 'svn diff' and 'svn merge' on the kill_parrot_conf branch; 
then the problem reoccurred.  During that window, I noted that there 
were 21 files in conflict -- which seems on the high side for us.

I don't know how the kill_parrot_conf was merged into trunk; that 
happened past my bedtime.  I do know that I was able to create a *new* 
branch by 'svn cp'-ing the kill_parrot_conf branch, and then was able to 
perform 'svn diff' of that branch against trunk (at old branch's 
fork-point) and 'svn merge' into sandbox without incident.  So my hunch 
is that this is something specific to a particular branch.

I see that we have a new cert.  It will be interesting to see whether 
these problems recur when we try to merge other branches created 
pre-cert-expiration into trunk post-new-cert.

Thank you very much.
kid51