Re: comment file naming

On Dec 28, 2004, at 10:45 PM, Bill Mill wrote:

> I want to change the way that comments are named, but I also want to
> solicit feedback from others before I do. All this comes about because
> I'm writing a comment administrator plugin for pyblosxom. (it's not
> yet in a usable state.)
>
> Currently, comments are named "<file comment attached to minus
> extension>-<time of comment file writing>.<comment extension>". This
> has several drawbacks. First, 2 comments on a single story cannot be
> made at the same time (imagine a dual-processor machine, or 2 machines
> writing to a shared storage medium) without undefined results. Yes, it
> is unlikely, but certainly not impossible.
>
> Second, the date in the file name is used to sort the file entries.
> While this is convenient, it puts significant meta-data into the
> filename. Thus, changing the time of a post in such a way that
> comments are correctly sorted involves renaming the comment whose time
> has changed. Furthermore, the user is responsible for ensuring that no
> two comments are set to have the same time, which seems a silly
> requirement.

It's probably faster/smarter to sort by mtime, if you're going to do it 
by filesystem metadata.

> So, if you agree that the naming scheme should be changed, how should
> it be changed? I will be glad to code up a patch, but I wanted to ask
> for suggestions on how comments could be better named - preferably in
> such a way as to not break backwards compatibility. (This would
> require changing the comment sorting to use the pubDate inside the
> file, but I think this should be done anyway).
>
> My initial thought is "<file comment attached to minus
> extension>-<randint(1, 100000)>.<comment extension>". The comments
> plugin would simply have to assure that the file doesn't already
> exist. Does anybody have a better naming scheme? Comments on this one?
> Should the number be significant?

I don't see how this is better than using a timestamp.. you can resolve 
conflicts in timestamp just like you can resolve conflicts in random 
numbers - try something else.  The tempfile module is better suited for 
what you want to do, it already has the machinery to create a unique 
file with a given prefix/suffix in some directory.  There are other 
methods, but tempfile seems like the One Obvious Way.

Personally I would want to put comments in a directory if you're 
reorganizing things.  Using a prefix on the name is kinda weird when 
you ONLY care about comments in the context of a particular entry, no 
need to search a global "commentspace" when you could have a per-entry 
"commentspace" just as easily.

As for your reply regarding string encoding, I only have one thing to 
say: **U N I C O D E**.  utf-8 would be the obvious choice.

-bob



-------------------------------------------------------
SF email is sponsored by - The IT Product Guide
Read honest & candid reviews on hundreds of IT Products from real users.
Discover which products truly live up to the hype. Start reading now. 
http://productguide.itmanagersjournal.com/