roller-commits mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Anil Gangolli (JIRA)" <nore...@atlassian.com>
Subject [Roller-JIRA] Commented: (ROL-1145) Roller calendar is spider trap
Date Fri, 20 Apr 2007 15:19:55 GMT

    [ http://opensource.atlassian.com/projects/roller/browse/ROL-1145?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#action_13701
] 

Anil Gangolli commented on ROL-1145:
------------------------------------

Attaching history from the e-mail thread surrounding the rollback of the earlier fix and the
reasons for it.  At last juncture the thread suggests this was fixed by Dave in the trunk,
but I still need to verify.


--------------------------------------------
Thread is In reverse chronological order...

Anil Gangolli <anil@busybuddha.org> wrote:

OK.

Allen had already noted that weblog.dateCreated tends to be inaccurate.

I had considered something along the lines of your approach in (2) but 
discarded that due to worries about performance impact.  I'm curious how you 
approached this.

I was considering an alternative where we would maintain the earliest ever 
publish date (the earliest publish date ever written --even it if is moved 
forward or deleted later) but sticking with the existing "strict prev: 
approach.

I'm fine with backing out for 3.1 so we have more time to determine the 
right approach for 3.2.

--a.




----- Original Message ----- 
From: "Dave" <snoopdave@gmail.com>
To: <roller-dev@incubator.apache.org>
Sent: Tuesday, December 12, 2006 1:48 PM
Subject: Re: svn commit: r482704 - 
/incubator/roller/branches/roller_3.1/src/org/apache/roller/ui/core/tags/calendar/CalendarTag.java


> Actually, my fix requires more extensive changes than I'd like to add
> to 3.1 so what I'm planning to do is to back out the r482704 change
> for 3.1 and commit my code so we can have a fix for 3.2. I'll start by
> committing my code to trunk.
>
> - Dave
>
>
>
> On 12/12/06, Dave <snoopdave@gmail.com> wrote:
>> Thanks for this important fix Anil, but there are two small problems:
>>
>> 1) I've found that weblog.dateCreated is not very reliable. Since it
>> is a timestamp, it gets updated anytime somebody makes a change to the
>> WEBSITE table via SQL UPDATE (unless they are  careful to use
>> updatetime=updatetime). I know our dateCreated information is
>> unreliable because of that. We rolled out the "3.2" codebase recently
>> and got some bug reports because of this.
>>
>> 2) Users don't want to click prev, prev, prev to go hunting back in
>> time for the previous month of entries. We don't just want to link to
>> the chronologically previous and previous months, we want to link to
>> the next and previous non-empty months. That's better for usability
>> and spidering reduction too.
>>
>> I have a fix ready to commit that addresses both of these issues.
>>
>> - Dave
>>
>>
>>
>> On 12/5/06, gangolli@apache.org <gangolli@apache.org> wrote:
>> > Author: gangolli
>> > Date: Tue Dec  5 08:37:49 2006
>> > New Revision: 482704
>> >
>> > URL: http://svn.apache.org/viewvc?view=rev&rev=482704
>> > Log:
>> > Fix for ROL-1312 integrated from the mainline.
>> >
>> > Modified:
>> > 
>> > incubator/roller/branches/roller_3.1/src/org/apache/roller/ui/core/tags/calendar/CalendarTag.java
>> >
>> > Modified: 
>> > incubator/roller/branches/roller_3.1/src/org/apache/roller/ui/core/tags/calendar/CalendarTag.java
>> > URL: 
>> > http://svn.apache.org/viewvc/incubator/roller/branches/roller_3.1/src/org/apache/roller/ui/core/tags/calendar/CalendarTag.java?view=diff&rev=482704&r1=482703&r2=482704
>> > ==============================================================================
>> > ---  
>> > incubator/roller/branches/roller_3.1/src/org/apache/roller/ui/core/tags/calendar/CalendarTag.java

>> > (original)
>> > +++ 
>> > incubator/roller/branches/roller_3.1/src/org/apache/roller/ui/core/tags/calendar/CalendarTag.java

>> > Tue Dec  5 08:37:49 2006
>> > @@ -163,8 +163,9 @@
>> >              day = DateUtil.getNoonOfDay(day, cal);
>> >              cal.set( Calendar.DAY_OF_MONTH, 
>> > cal.getMinimum(Calendar.DAY_OF_MONTH) );
>> >
>> > -            // go back to sunday before that: the first sunday in the 
>> > calendar
>> > -            while ( cal.get( Calendar.DAY_OF_WEEK ) != 
>> > Calendar.SUNDAY ) {
>> > +            // Go back to first day of week before that (Sunday in US, 
>> > Monday in France, e.g.)
>> > +            // in the calendar
>> > +            while ( cal.get( Calendar.DAY_OF_WEEK ) != 
>> > cal.getFirstDayOfWeek() ) {
>> >                  cal.add( Calendar.DATE, -1 );
>> >              }
>> >
>> >
>> >
>> >
>>

> Roller calendar is spider trap
> ------------------------------
>
>                 Key: ROL-1145
>                 URL: http://opensource.atlassian.com/projects/roller/browse/ROL-1145
>             Project: Roller
>          Issue Type: Bug
>          Components: Macros
>    Affects Versions: 2.1
>            Reporter: Anil Gangolli
>            Assignee: Anil Gangolli
>            Priority: Minor
>             Fix For: 3.2
>
>
> Filed based on the following e-mail message.
> ----- Original Message ----- 
> From: "Trygve Lie" <trygvelie@hotmail.com>
> To: <roller-dev@incubator.apache.org>
> Sent: Monday, May 22, 2006 3:55 AM
> Subject: Spider trap in Roller's calendar
> > Hi
> > 
> > The calendar in Roller can cause a small problem for search engine spiders 
> > since it's possible to page backwards in dates by the calendar. It's 
> > actually possible to page pack to the year zero...
> > Ex: http://rollerweblogger.org/page/roller/000104
> > 
> > A spider hitting such a "trap" will just continue to page backwards until it 
> > "gets tired".
> > 
> > There are two dangerous problems with this:
> > - This can cause unnecessary stress on the server running Roller (ex; Yahoos 
> > spider make big slurps and do actually not consider if the server can handle 
> > it or not).
> > - At some point the spider will "get tired" because such paging will 
> > generate a lot of similar pages (when there are no content all pages will be 
> > similar) and the spider will then mark the site as "possible spam" due to 
> > all the similar pages.
> > 
> > I would like to suggest that there might be added a small check which makes 
> > the backward paging in the calendar only go back to when the first post was 
> > made in the blog. This would cause the calendar to page back to the month 
> > when the first post was added to the blog. To be able to page beyond that 
> > month does not have any actual interest.
> > 
> > Kind regards
> > Trygve Lie
> > 
> > _________________________________________________________________
> > MSN Spaces http://spaces.msn.com/?mkt=nb-no Vis hvem du er og hva du vil
> >

-- 
This message is automatically generated by JIRA.
-
If you think it was sent incorrectly contact one of the administrators: http://opensource.atlassian.com/projects/roller/secure/Administrators.jspa
-
For more information on JIRA, see: http://www.atlassian.com/software/jira

        

Mime
View raw message