pretxnchangegroup hook fails because it can't acquire lock

[Mick Jordan]

On 11/5/13 1:40 PM, Simon King wrote:
> On Tue, Nov 5, 2013 at 7:50 PM, Mick Jordan <mick.jordan at oracle.com> wrote:
>> On 11/5/13 3:41 AM, Simon King wrote:
>>> [hooks] pretxnchangegroup.push =
>>> python:/path/to/hook/script.py:pushwhilelocked
>>
>>> Hope that helps, Simon
>> I have this working now, just one small problem that someone might know the
>> answer to, which is that the hgweb script is evidently returning a bad
>> header. It's nice that all the python logging shows up now in the server
>> error_log Here are the salient lines:
>>
>> [Tue Nov 05 11:31:25 2013] [error] [client ...] abort:
>> pretxnchangegroup.run_gate hook failed
>> [Tue Nov 05 11:31:25 2013] [error] [client ...] malformed header from
>> script. Bad header=listing keys for "bookmarks": hgweb.cgi
>>
>>
> At a guess, your hook is producing some output on stdout. Since you
> are using CGI, stdout from the script becomes part of the HTTP
> response.
Indeed, rather a lot I am afraid. But it's useful debugging info, so is 
there a correct way to do that which doesn't become part of the response?
>
> So you need to prevent your hook from outputting to stdout. If it were
> an out-of-process hook, you could do this using standard shell
> redirection operators, but for an in-process hook you'll have to do
> something different. Completely untested, but you might be able to do
> something like this:
>
> import sys
>
> def myhook(ui, repo, **kwargs):
>      oldstdout = sys.stdout
>      try:
>          sys.stdout = sys.stderr
>          do_stuff()
>      finally:
>          sys.stdout = oldstdout
>
>
> If you have the option, it might be better to use WSGI rather than CGI
> to host hgweb. The performance should be better, and it wouldn't have
> problems like this.
>
>
I looked at WSGI this morning, but I have to build it from source for my 
version of Linux/Apache. I'll get to that, but why would it avoid the 
output problem?

Thanks
Mick