RFC: Managing Mercurial Repositories Remotely

Jesse Glick jesse.glick at sun.com
Wed Feb 20 15:57:46 CST 2008


Peter Arrenbrecht wrote:
>>> The reason I wanted to support MQ on the server was for allowing 
>>> easy web review of patches maintained in a queue.
>> 
>> Why can't you just link to the actual patch (in the queue
>> repository as browsable by hgwebdir)?
> 
> That does not give me annotate or, if hgweb ever supports it, diffs 
> with more context on the patched files [...for use in] reviewboard.

I still don't get it. Running MQ commands on a remote repository seems 
like a terrible idea to me. If you just have your "real" remote 
repository which people develop as usual, and a remote patch queue 
repository which people can also push into, then all your information is 
available.

If you want Reviewboard or any other web-based tool to be able to 
display a patch in the context of unmodified lines or files, it can 
simply maintain a private local clone of both repositories and use 
regular local MQ commands to calculate the appearance of patches in 
context. Such a server could decide either to have one MQ-managed local 
repo and serve only one web request at a time (running e.g. 'hg qgo && 
hg qdi'), or maintain a pool of such repos to permit concurrent display 
of different patches, or run a background process to push patches and 
cache displayable information resulting from them, ad nauseam - none of 
which would be safe or efficient if it had to make remote calls.



More information about the Mercurial-devel mailing list