I will fix this in today's afternoon.... sorry about that.

(I have been away for 2 weeks and just getting back to work)

On Thu, Jul 12, 2018 at 10:09 AM, Robbie Gemmell
<robbie.gemm...@gmail.com> wrote:
> The 2.6.1 docs link at
> http://activemq.apache.org/artemis/past-releases.html is broken. The
> link itself seems fine this time and the other 2.6.1 links work, there
> just dont seem to be any 2.6.1 docs.
>
> The 2.6.0 and 2.6.1 releases aren't mentioned on
> http://activemq.apache.org/artemis/previous-docs.html
>
> Robbie
>
> On 20 June 2018 at 13:20, Robbie Gemmell <robbie.gemm...@gmail.com> wrote:
>> As I said some links just weren't updated to reflect the newer
>> checksum distribution policy, but most were broken from pointing at
>> the wrong download area/server and using the query arguments that
>> aren't used there. That suggested the content was copied unchanged
>> from download.html into past-releases.html, which it generally
>> shouldnt be.
>>
>> This commit 
>> https://lists.apache.org/thread.html/5e07407b600aaae38a7ed757a0008bc70cca7a712868812202189f28@%3Ccommits.activemq.apache.org%3E
>> shows a couple of issues:
>> - All of the links for 2.5.0 and 2.6.0 were broken except for the
>> release notes and commit report links. The main links pointed to the
>> wrong download server/area, with the tar/zip links also having query
>> args left from the mirror system. The doc links pointed to the central
>> docs page rather than the actual docs for the release.
>> - The 1.5.6 release entry had links pointing to .sha1 files that dont
>> exist. Those were presumably just missed in updating the links from
>> 1.5.5, which did have .sha1 checksums. I updated those to link to the
>> .sha512 files from the current distribution policy.
>>
>> The typo you mention below was trivial so I didnt call it out in the
>> mail but adjusted that in this commit:
>> https://lists.apache.org/thread.html/3d6c90c59e30d65013a0f5bc7243434edb8eb817ffd2272b3be341de@%3Ccommits.activemq.apache.org%3E
>>
>> A final change is mostly just reducing indent, as I removed the
>> wrapper divs (presumably also copied from download.html) for the
>> latest entries since the earlier ones dont have those and they all
>> live within an outer wrapper div already:
>> https://lists.apache.org/thread.html/641671780a10b81419711805519e7a8643688be53705c844ab7520de@%3Ccommits.activemq.apache.org%3E
>>
>> On 19 June 2018 at 15:59, Clebert Suconic <clebert.suco...@gmail.com> wrote:
>>> I will be more careful... although.. beyond a typo on the
>>> description.. of 2.6.0 in one of the downloads.. I had something like
>>> <a href = "....2.6.0..."> 2.5.0 </a> what was the actual mistake.
>>>
>>>
>>> I'm a bit rusty in svn.. and even though I could find the diff.. I did
>>> not understand what was my actual mistake.
>>>
>>> On Tue, Jun 19, 2018 at 6:37 AM, Robbie Gemmell
>>> <robbie.gemm...@gmail.com> wrote:
>>>> I tried downloading some Artemis bits using links from the website
>>>> earlier and found various links not working. Some just weren't updated
>>>> to reflect the newer checksum distribution policy, but most were
>>>> broken from pointing at the wrong download area/server and using the
>>>> query arguments that no longer apply. I've fixed the ones I saw now,
>>>> but as I had also done the same for earlier releases previously I
>>>> think its now worth noting the issue here to hopefully avoid future
>>>> breakages occurring.
>>>>
>>>> It looks like the content is being copied unchanged from the main
>>>> download page to the past releases page. This wont work because the
>>>> main page links to the download mirrors and that content is removed
>>>> for historic releases as newer ones occur, so any links for past
>>>> releases must reference the archive. The mirrored download links also
>>>> use query arguments that don't apply for the archive area.
>>>>
>>>> Instead, I think the simplest thing to do is to copy the prior release
>>>> entry on the past releases page and update the version numbers with
>>>> basic find/replace. I went to update the RELEASING.md file to suggest
>>>> this but found it already says that, so thought I'd send a mail to
>>>> highlight the issue instead.
>>>>
>>>> Robbie
>>>
>>>
>>>
>>> --
>>> Clebert Suconic



-- 
Clebert Suconic

Reply via email to