[isabelle-dev] AFP sitegen

[Christian Sternagel]

Thanks for the quick answers. In case I'm not supposed to run 
tool/sitegen myself, just let me know ;)

cheers

chris

On 06/06/2013 03:54 PM, Andreas Lochbihler wrote:
> Sorry for the confusion, I never ran sitegen.py myself because I thought
> that to be the priviledge of the editors. As Gerwin has found out, I
> dropped these links manually in 376347e6131a because they all were
> broken after the update on sourceforge. I decided not to update them for
> three reasons:
>
> 1. Most other entries do not link to changeset revisions; Category is
> now the only exception. Often they don't even mention the revision ID at
> all.
>
> 2. It is unclear when the links will break again if they are not checked
> automatically.
>
> 3. On the entry page, the links are visually the most prominent part of
> the change history, although they are the least relevant bit of
> information in the change history.
>
> If sitegen.py automatically links to the changesets, I'd be happy. But
> then, I suggest that the links are not formatted as highlighted as they
> are now.
>
> Andreas
>
> On 06/06/13 08:09, Gerwin Klein wrote:
>> It looks like Andreas dropped these manually for his entries, so
>> nothing really went wrong with the tools, he was just reacting to the
>> sourceforge update leading to broken links.
>>
>> The URL scheme for linking to revision IDs in the new sourceforge
>> setup is
>>
>> http://sourceforge.net/p/afp/code/ci/change-set-hash
>>
>> The short hashes that we normally use seem to work fine (it shows you
>> long ones by default when you browse).
>>
>> It's up to the authors to have change set ids as links or not, so I'm
>> not adding them back in myself. If Andreas is reading this and prefers
>> having them in, by all means put them back.
>>
>> We haven't really made up our minds if developers should run
>> admin/sitegen after updating history in metadata. I'd say, if you feel
>> comfortable using sitegen and check that your changes are confined to
>> history (as Chris apparently did), this is Ok to do. If you're not
>> feeling comfortable doing this yourself, you change will just show up
>> on the devel website the next time someone runs sitegen.
>>
>> We could try make sitegen.py aware of hg revision ids and make it link
>> them automatically. If there's a volunteer for implementing this, I'm
>> happy to consider this.
>>
>> Cheers,
>> Gerwin
>>
>> On 06.06.2013, at 1:52 PM, Gerwin Klein <gerwin.klein at nicta.com.au>
>> wrote:
>>
>>> I'll have a look at it. The links shouldn't be dropped, something is
>>> going wrong there.
>>>
>>> Cheers,
>>> Gerwin
>>>
>>> On 06/06/2013, at 1:48 PM, Christian Sternagel
>>> <c.sternagel at gmail.com> wrote:
>>>
>>>> Btw: the links do not seem to work anyway. But why not replace them
>>>> with working links instead of just dropping them?
>>>>
>>>> On 06/06/2013 12:40 PM, Christian Sternagel wrote:
>>>>> Dear all,
>>>>>
>>>>> to update the change history of one of my AFP entries, I ran
>>>>> admin/sitegen. I noticed that as a result some other sites changed
>>>>> too.
>>>>> All the changes where along the lines of
>>>>>
>>>>> -(revision <a
>>>>> href="http://afp.hg.sourceforge.net/hgweb/afp/afp/rev/f74a8be156a7">f74a8be156a7</a>)<br>
>>>>>
>>>>>
>>>>> +(revision f74a8be156a7)<br>
>>>>>
>>>>> in corresponding *.shtml files, i.e., links to changesets are replaced
>>>>> by the mere short-form changeset ID. Is this on purpose or did I do
>>>>> something wrong? (I will of course refrain from pushing any changes
>>>>> until I got an answer.)
>>>>>
>>>>> cheers
>>>>>
>>>>> chris
>>>>