incubator-ooo-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Kay Schenk <>
Subject Re: [UPDATE SERVICE] proposal for a AOO 3.4 update service
Date Wed, 16 May 2012 17:38:11 GMT
On Tue, May 15, 2012 at 11:37 PM, Oliver-Rainer Wittmann <> wrote:

> Hi,
> On 15.05.2012 20:45, Oliver-Rainer Wittmann wrote:
>> Hi,
>> Am 15.05.12 16:11, schrieb Rob Weir:
>>> On Tue, May 15, 2012 at 8:11 AM, Oliver-Rainer Wittmann
>>> <> wrote:
>>>> Hi,
>>>> From my point of view it would make sense to reactivate a simple update
>>>> service for AOO 3.4.
>>>> The update URL for AOO 3.4 is:
>>>> **Update<>
>>>> (plus a query part ?pkgfmt=<pkgformat> for non-Windows platforms)
>>>> As this URL resolves to nothing, the user currently gets the following
>>>> response from the update functionality in AOO 3.4:
>>>> Status: Checking for an update failed.
>>>> Description: General Internet error has occurred.
>>>> I propose provide the following XML document when a HTTP GET request to
>>>> the
>>>> above given URL is made:
>>>> <?xml version="1.0" encoding="UTF-8"?>
>>>> <inst:description
>>>> xmlns:inst="http://****description<>
>>>> ">
>>>> </inst:description>
>>>> Kay already made such an XML document available at:
>>>> This response would allow the update functionality in AOO 3.4 to return
>>>> to
>>>> the user that the version is up to date.
>>>> Thus, to reactivate an working update service for AOO 3.4 a redirection
>>>> is
>>>> needed.
>>> Are proposing that we just have a static XML file and redirect the
>>> requests so it loads that static file?
>> Yes, as a short-term and fast solution.
>>  I can see that as being a useful short-term solution. But soon we'll
>>> need some more complicated logic, right? For example, when we enable
>>> the 3.3. update check, we'll need to know that updates are available
>>> for some languages, but not others. Can we do that all with
>>> redirection to static files? Or do we need server-based logic, i.e.,
>>> a cgi script?
>> Static files would be possible, because each version has its own update
>> service
>> URL, but it would be not the best solution for the long-term.
>> Thus, some server-based logic would make sense.
>>  If we're going to need a cgi script in the end, I wonder if it makes
>>> sense to start with one now? We could have a simply script that today
>>> just always points to the "no update available" XML for AOO 3.4. But
>>> then we make it more complicated as we go.
>> I am currently in preparation of a proposal for an update service for OOo
>> 3.3
>> installations. Here, I can/will demonstrate how a server-based logic
>> would look
>> like.
>>  Can somebody make this happen?
>>>> I have to admit that do not have the knowledge to do it on my own.
>>> If we just redirect to a static file, I think you can just enter a
>>> JIRA request for Infra. If we go with a cgi script then we need
>>> someone to develop that script first.
>> If nobody objects, I would go for this short-term and static approach and
>> would
>> ask via JIRA request for Infra, if the redirect to the already existing
>> static
>> file can be established.
> I will use issue 119361 -**
> show_bug.cgi?id=119361<>- to
track the progress on this task.
> Best regards, Oliver.

OK, and a very short update on this.

I tried to deal with this and continually ran into issues.

At the simplest, I tired to make up a "generic" update for maybe all
platforms and languages that would just take them to a page to choose an
update -- basically our download/other.html at this point.

However, if you exclude the platform and other particulars in a very simple
XML file, nothing happens -- in other words, the URL is just ignored and
you get a "no updates" message. This is what is in:

Also, life is complicated by appending the "pkgfmt " on update strings in

<AOO install directory>/program/versionrc (for will vary
depending on OS)

Finally, whatever we do...assuming we can get a feed to work...under NO
circumstance should we redirect to the
"dummy" host. The older version 3.0 did something entirely different, and
when I had infra do this redirect some months ago, it caused the Apache
webserver complex to crash.

I *thing* the newer services update32, update33, update35, etc. will be
fine, but we need to verify this.


"Well, life has a funny way of sneaking up on you
 And life has a funny way of helping you out
 Helping you out."
                            -- "Ironic", Alanis Morissette

  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message