Re: [pgsql-www] Excessive # usage in URLs

On Fri, Oct 23, 2015 at 5:28 PM, Magnus Hagander <magnus(at)hagander(dot)net>
wrote:

>
>
> On Fri, Oct 23, 2015 at 5:26 PM, Alvaro Herrera <alvherre(at)2ndquadrant(dot)com>
> wrote:
>
>> Stephen Frost wrote:
>> > * Magnus Hagander (magnus(at)hagander(dot)net) wrote:
>> > > The point of the # is that you actually want to point to that specific
>> > > message in the thread. If you just pick the first message in the
>> thread we
>> > > could exclude it, but how else would you link to a specific message?
>> >
>> > The request isn't to remove the "#" ability for the flat view entirely,
>> > it'd just be nice if the site didn't redirect to having the "#" when the
>> > message-id after the "#" is the same as the one before the "#" (iow,
>> > it's the first message).
>>
>> Here's another idea. Right now the <A NAME="tag"> tag is the
>> message-id, but since messages are added consecutively to a thread, we
>> could just assign each of them a number, so that to link the second
>> message in a thread we could do this
>>
>> http://www.postgresql.org/message-id/flat/20151023125233(dot)GQ3685(at)tamriel(dot)
>> snowman.net#2
>> We could do this in addition to the message-id-based NAME tag.
>>
>
> True. We could also use the surrogate key we have in the db.
>
> In fact, we could have a <a name> for each of those two, so that one could
> still use the other one if building URLs elsewhere, but the ones generated
> in the archives could use the shorter id. That's a number that goes up to
> max 6 digits at this point (and that would actually be guaranteed to be
> stable in the future, unlike just using the sequential)
>

So, do I win anything for best zombie thread back from the dead today?

It's only been a few years, so I've now actually done this in the shape of
two tiny commits and I think totaling 11 lines :)

Basically, now:
* If you navigate to the first message in a thread, no redirect at all is
done
* If you navigate to another message, there is a redirect to the # url but
using the md5 of the messageid keeping a lot shorter than your average
gmail messageid

Well, that's actually more than 4 lines of code per year, you can't expect
coding much faster than that!

FWIW, I haven't purged caches so it will gradually start showing up on
threads in the archives as time go by.

--
Magnus Hagander
Me: https://www.hagander.net/ <http://www.hagander.net/>
Work: https://www.redpill-linpro.com/ <http://www.redpill-linpro.com/>