Re: PATCH: add "current" version link to docs page - Mailing list pgsql-www

On 11/04/2015 10:06 PM, Amir Rohan wrote:
> On 11/04/2015 09:17 PM, Stefan Kaltenbrunner wrote:
>> On 10/21/2015 06:45 PM, Amir Rohan wrote:
>>> On 10/21/2015 05:04 PM, Alvaro Herrera wrote:
>>>> Amir Rohan wrote:
>>>>> On 10/05/2015 01:57 PM, Amir Rohan wrote:
>>>>>> I previously suggested this could help SEO:
>>>>>>
>>>>>> http://www.postgresql.org/message-id/560614CA.1080304@mail.com
>>>>>>
>>>>>> There wasn't much of a reaction but hopefully actually providing
>>>>>> the patch might make the difference.
>>>>>>
>>>>>> You'll now see something like:
>>>>>>
>>>>>> 9.2 / 9.3 / current (9.4)
>>>>>>
>>>>>> at the top of the page, with "current" linking /docs/current
>>>>>> and "9.4" linking /docs/9.4.
>>>>>>
>>>>>> For SEO purposes I think it would actually be better to elide
>>>>>> the numbered-version altogether, so people will have to do extra work
>>>>>> *not* to link to /docs/current/, but I figured that would turn into
>>>>>> a bikeshed, so this less intrusive version is what I'm sending.
>>>>>
>>>>> Bump. There's been no response to this, does the silence imply rejection
>>>>> or just very low priority?
>>>>
>>>> I think this is a good idea, and obviously so does Dave.
>>>>
>>>
>>> I'll bump periodically until it goes live or someone tells me to stop...
>>
>> Sorry for this taking so long - but I now have applied PATCH 2/2, I
>> didnt bother applying the fixtures one due to multiple issues (some
>> preexisting):
>>
>> * added hunk in the patch clearly was confused about 9.0 vs 8.0
>> * random whitespace added to some lines
>> * the entire current fixture file is kinda outdated wrt the current data
>> model and needs much bigger rejiggering...
>>
>>
>> thanks for the patch!
>>
>>
>> Stefan
>>
> 
> Just to set things straight:
> 
> - The trailing space wasn't added, it was copy-pasted from the rest of
> the file which has it everywhere.

heh - doesnt make it any better :)

> 
> - I wasn't confused about 8.0 vs. 9.0, the fixture is simply *that* old.
> It doesn't matter if the data is "correct" (and it won't be, in a year
> or 5), but it needs datums that cover the data model or you can't test
> your changes.

well citing a part of the diff:


+  "pk": 6,
+  "model": "core.version",
+  "fields": {
+   "relnotes": "release.html#RELEASE-8-0-21",
+   "tree": "9.0",
+   "testing" : 1,
+   "supported": false,
+   "reldate": "2009-03-17",
+   "eoldate": "2010-10-01",
+   "firstreldate": "2009-03-17",
+   "latestminor": 21


"tree" -> "9.0" and "relnotes" -> "release.html#RELEASE-8-0-21 (as well
as "latestminor" -> "21" feel completely wrong in that combination :)


> 
> Guess why I was apprehensive about asking for an update and waiting for
> someone to find the time. :)
> 
> - Yes, the whole fixture file is really hopelessly out of date and
> useless for testing, so If you're going to update it with live data
> , great, but if not please reconsider the patch - I spent about as much
> time fixing things so I could test my (trivial) change as the change itself.

I'm actually working on a more complete fix to the fixture file now...


Stefan