fofredux-devel

[Fofredux-devel] cache_time on feeds

[Evan R. <eva...@gm...>]

hey guys,

a side note i just remembered.

in the feed items table, on my install, i've stripped out the
auto-timestamping on the cache_time field...since this is actually changing
valid data from the rss feed.  in order to do it correctly, i'd add another
field for date_changed or something, which is then updated when the user
marks the item read (or whatever other action).

any objections to me correcting that, or shall i just add it as a
bug/feature request for the time-being?  (if sourceforge wants to work...i
am getting only 500 errors tonight.)

/evan

Re: [Fofredux-devel] cache_time on feeds

[Kevin <ke...@dr...>]

> hey guys,
>
> a side note i just remembered.
>
> in the feed items table, on my install, i've stripped out the
> auto-timestamping on the cache_time field...since this is actually
> changing
> valid data from the rss feed.  in order to do it correctly, i'd add
> another
> field for date_changed or something, which is then updated when the use=
r
> marks the item read (or whatever other action).
>
> any objections to me correcting that, or shall i just add it as a
> bug/feature request for the time-being?  (if sourceforge wants to work.=
..i
> am getting only 500 errors tonight.)

publish_time uses the timestamp from the rss feed.
cache_time is when it is inserted into the database.  It is primarily use=
d
for items deletes.  We also display it in the view.

Do we care when a user marks an item read?  How is that useful?

--=20
Kevin

Re: [Fofredux-devel] cache_time on feeds

[Evan R. <eva...@gm...>]

well, i never delete anything.  so for me it's useful to have an idea when =
i
actually read something.  and when it was cached is seperate data that
shouldn't be tampered with.  it's your basic principle of having an insert
and update timestamp on the record.
in the end it's not absolutely necessary, but the auto-timestamping on
cache_time needs to go.  i suspect that the adodb stuff adds it, when we ad=
d
the field, as it's the first timestamp in the table.  i don't want the
cache_time to be changed when i mark something read.  (if we don't know whe=
n
it was marked read, then it's not a big problem right now)


On 4/21/06, Kevin <ke...@dr...> wrote:
>
>
> > hey guys,
> >
> > a side note i just remembered.
> >
> > in the feed items table, on my install, i've stripped out the
> > auto-timestamping on the cache_time field...since this is actually
> > changing
> > valid data from the rss feed.  in order to do it correctly, i'd add
> > another
> > field for date_changed or something, which is then updated when the use=
r
> > marks the item read (or whatever other action).
> >
> > any objections to me correcting that, or shall i just add it as a
> > bug/feature request for the time-being?  (if sourceforge wants to
> work...i
> > am getting only 500 errors tonight.)
>
> publish_time uses the timestamp from the rss feed.
> cache_time is when it is inserted into the database.  It is primarily use=
d
> for items deletes.  We also display it in the view.
>
> Do we care when a user marks an item read?  How is that useful?
>
> --
> Kevin
>
>
>
> -------------------------------------------------------
> Using Tomcat but need to do more? Need to support web services, security?
> Get stuff done quickly with pre-integrated technology to make your job
> easier
> Download IBM WebSphere Application Server v.1.0.1 based on Apache Geronim=
o
> http://sel.as-us.falkag.net/sel?cmdlnk&kid=120709&bid&3057&dat=121642
> _______________________________________________
> Fofredux-devel mailing list
> Fof...@li...
> https://lists.sourceforge.net/lists/listinfo/fofredux-devel
>

Re: [Fofredux-devel] cache_time on feeds

[Kevin <ke...@dr...>]

Evan Roth wrote:
...
> in the end it's not absolutely necessary, but the auto-timestamping on 
> cache_time needs to go.  i suspect that the adodb stuff adds it, when 
> we add the field, as it's the first timestamp in the table.
I still don't understand.   What about the handling of the cache_time 
field is wrong?   Could you provide a test case so I can reproduce it?

-Kevin

Re: [Fofredux-devel] cache_time on feeds

[Evan R. <eva...@gm...>]

The cache_time has an attribute of 'ON UPDATE CURRENT_TIMESTAMP'.  This is
obviously new in redux, since the original FoF did not use cache_time.  i
was doing some tests last night, and it seems that the first TIMESTAMP fiel=
d
in a table gets this attribute automagically from ADODB (which was also not
included in FoF, iirc).

when you are reading an item, it is ok, since the field value is correct.
when you mark it read or save or do anything else to the record, this
attribute triggers, and cache_time is rewritten to the time of the update.
Thus, we lose the information when we stored this bit of data.

If you ask me, i'd be of the opinion that we need 3 dates:
when was it published -> taken from RSS
when was it cached -> on insert
when was it changed -> on update (ie. read, unread, saved, etc...)

one could argue that the cache_time is not important, since we have publish
time...but often they are largely different for me:
1.  being in germany, and reading mostly stuff from the homeland...i am
dealing with 6-9 hrs difference, which could not come through.  ie.
published at 12:00 local time EST, but cached at 18:00 local time CET.
2.  and the most important point for me...often times there is a delay from
when content is published to when it shows up in an rss feed.  hence, we ge=
t
a new item 2 hrs after it was really published, and therefore, it won't sho=
w
on the top of the list.

i'd prefer to be able to sort my items by cache_time, rather than
publish_time, to see the new items (to me!) always at the top.

On 4/21/06, Kevin <ke...@dr...> wrote:
>
> Evan Roth wrote:
> ...
> > in the end it's not absolutely necessary, but the auto-timestamping on
> > cache_time needs to go.  i suspect that the adodb stuff adds it, when
> > we add the field, as it's the first timestamp in the table.
> I still don't understand.   What about the handling of the cache_time
> field is wrong?   Could you provide a test case so I can reproduce it?
>
> -Kevin
>
>
> -------------------------------------------------------
> Using Tomcat but need to do more? Need to support web services, security?
> Get stuff done quickly with pre-integrated technology to make your job
> easier
> Download IBM WebSphere Application Server v.1.0.1 based on Apache Geronim=
o
> http://sel.as-us.falkag.net/sel?cmd=3Dlnk&kid=3D120709&bid=3D263057&dat=
=3D121642
> _______________________________________________
> Fofredux-devel mailing list
> Fof...@li...
> https://lists.sourceforge.net/lists/listinfo/fofredux-devel
>

Re: [Fofredux-devel] cache_time on feeds

[Kevin <ke...@dr...>]

Evan Roth wrote:
> The cache_time has an attribute of 'ON UPDATE CURRENT_TIMESTAMP'.  
> This is obviously new in redux, since the original FoF did not use 
> cache_time.  i was doing some tests last night, and it seems that the 
> first TIMESTAMP field in a table gets this attribute automagically 
> from ADODB (which was also not included in FoF, iirc).

This is mysql doing it, not ADODB_Lite.   If you enable FOF_DB_DEBUG,  
you can see all the SQL sent to the database. I can reproduce this with 
Mysql 4.1.11.  The first timestamp column added to a table gets this 
attribute set.(ON UPDATE CURRENT_TIMESTAMP)  If I reverse publish_time 
and cache_time in the table create statement, then publish_time get's 
this *feature*, instead of cache_time.

Have you created a bug for this yet?   I'll add it if I have time 
today.**  

<...>
> one could argue that the cache_time is not important, since we have 
> publish time...but often they are largely different for me:
> 1.  being in germany, and reading mostly stuff from the homeland...i 
> am dealing with 6-9 hrs difference, which could not come through.  
> ie.  published at 12:00 local time EST, but cached at 18:00 local time 
> CET.
We convert all timestamps to GMT internally. (well, unix epoch seconds)  
which should eliminate timezone problems, at least in the DB.  You can 
configure what your local time offset is on the options page.   This is 
used when timestamps are displayed on the view page.  We should probably 
make this smarter by allowing a timezone string to be entered instead of 
an hour offset value.
> 2.  and the most important point for me...often times there is a delay 
> from when content is published to when it shows up in an rss feed.  
> hence, we get a new item 2 hrs after it was really published, and 
> therefore, it won't show on the top of the list.
>
> i'd prefer to be able to sort my items by cache_time, rather than 
> publish_time, to see the new items (to me!) always at the top.
>
Allowing the sort column to be configurable is on the feature list, I think.

-Kevin

Re: [Fofredux-devel] cache_time on feeds

[Kevin <ke...@dr...>]

added bug #1477023.

> Evan Roth wrote:
>> The cache_time has an attribute of 'ON UPDATE CURRENT_TIMESTAMP'.
>> This is obviously new in redux, since the original FoF did not use
>> cache_time.  i was doing some tests last night, and it seems that the
>> first TIMESTAMP field in a table gets this attribute automagically
>> from ADODB (which was also not included in FoF, iirc).
>
> This is mysql doing it, not ADODB_Lite.   If you enable FOF_DB_DEBUG,
> you can see all the SQL sent to the database. I can reproduce this with
> Mysql 4.1.11.  The first timestamp column added to a table gets this
> attribute set.(ON UPDATE CURRENT_TIMESTAMP)  If I reverse publish_time
> and cache_time in the table create statement, then publish_time get's
> this *feature*, instead of cache_time.
>
> Have you created a bug for this yet?   I'll add it if I have time
> today.**
>
> <...>
>> one could argue that the cache_time is not important, since we have
>> publish time...but often they are largely different for me:
>> 1.  being in germany, and reading mostly stuff from the homeland...i
>> am dealing with 6-9 hrs difference, which could not come through.
>> ie.  published at 12:00 local time EST, but cached at 18:00 local time
>> CET.
> We convert all timestamps to GMT internally. (well, unix epoch seconds)
> which should eliminate timezone problems, at least in the DB.  You can
> configure what your local time offset is on the options page.   This is
> used when timestamps are displayed on the view page.  We should probabl=
y
> make this smarter by allowing a timezone string to be entered instead o=
f
> an hour offset value.
>> 2.  and the most important point for me...often times there is a delay
>> from when content is published to when it shows up in an rss feed.
>> hence, we get a new item 2 hrs after it was really published, and
>> therefore, it won't show on the top of the list.
>>
>> i'd prefer to be able to sort my items by cache_time, rather than
>> publish_time, to see the new items (to me!) always at the top.
>>
> Allowing the sort column to be configurable is on the feature list, I
> think.
>
> -Kevin
>
>
>
> -------------------------------------------------------
> Using Tomcat but need to do more? Need to support web services, securit=
y?
> Get stuff done quickly with pre-integrated technology to make your job
> easier
> Download IBM WebSphere Application Server v.1.0.1 based on Apache Geron=
imo
> http://sel.as-us.falkag.net/sel?cmd=3Dlnk&kid=3D120709&bid=3D263057&dat=
=3D121642
> _______________________________________________
> Fofredux-devel mailing list
> Fof...@li...
> https://lists.sourceforge.net/lists/listinfo/fofredux-devel
>


--=20
Kevin

Re: [Fofredux-devel] cache_time on feeds

[Evan R. <eva...@gm...>]

saw that...thanx.

do you have any idea how to eliminate this, i was playing with it last
week, and it seems pretty stubborn.

On 4/27/06, Kevin <ke...@dr...> wrote:
>
> added bug #1477023.
>
> > Evan Roth wrote:
> >> The cache_time has an attribute of 'ON UPDATE CURRENT_TIMESTAMP'.
> >> This is obviously new in redux, since the original FoF did not use
> >> cache_time.  i was doing some tests last night, and it seems that the
> >> first TIMESTAMP field in a table gets this attribute automagically
> >> from ADODB (which was also not included in FoF, iirc).
> >
> > This is mysql doing it, not ADODB_Lite.   If you enable FOF_DB_DEBUG,
> > you can see all the SQL sent to the database. I can reproduce this with
> > Mysql 4.1.11.  The first timestamp column added to a table gets this
> > attribute set.(ON UPDATE CURRENT_TIMESTAMP)  If I reverse publish_time
> > and cache_time in the table create statement, then publish_time get's
> > this *feature*, instead of cache_time.
> >
> > Have you created a bug for this yet?   I'll add it if I have time
> > today.**
> >
> > <...>
> >> one could argue that the cache_time is not important, since we have
> >> publish time...but often they are largely different for me:
> >> 1.  being in germany, and reading mostly stuff from the homeland...i
> >> am dealing with 6-9 hrs difference, which could not come through.
> >> ie.  published at 12:00 local time EST, but cached at 18:00 local time
> >> CET.
> > We convert all timestamps to GMT internally. (well, unix epoch seconds)
> > which should eliminate timezone problems, at least in the DB.  You can
> > configure what your local time offset is on the options page.   This is
> > used when timestamps are displayed on the view page.  We should probabl=
y
> > make this smarter by allowing a timezone string to be entered instead o=
f
> > an hour offset value.
> >> 2.  and the most important point for me...often times there is a delay
> >> from when content is published to when it shows up in an rss feed.
> >> hence, we get a new item 2 hrs after it was really published, and
> >> therefore, it won't show on the top of the list.
> >>
> >> i'd prefer to be able to sort my items by cache_time, rather than
> >> publish_time, to see the new items (to me!) always at the top.
> >>
> > Allowing the sort column to be configurable is on the feature list, I
> > think.
> >
> > -Kevin
> >
> >
> >
> > -------------------------------------------------------
> > Using Tomcat but need to do more? Need to support web services, securit=
y?
> > Get stuff done quickly with pre-integrated technology to make your job
> > easier
> > Download IBM WebSphere Application Server v.1.0.1 based on Apache Geron=
imo
> > http://sel.as-us.falkag.net/sel?cmd=3Dlnk&kid=3D120709&bid=3D263057&dat=
=3D121642
> > _______________________________________________
> > Fofredux-devel mailing list
> > Fof...@li...
> > https://lists.sourceforge.net/lists/listinfo/fofredux-devel
> >
>
>
> --
> Kevin
>
>
>
> -------------------------------------------------------
> Using Tomcat but need to do more? Need to support web services, security?
> Get stuff done quickly with pre-integrated technology to make your job ea=
sier
> Download IBM WebSphere Application Server v.1.0.1 based on Apache Geronim=
o
> http://sel.as-us.falkag.net/sel?cmdlnk&kid=120709&bid&3057&dat=121642
> _______________________________________________
> Fofredux-devel mailing list
> Fof...@li...
> https://lists.sourceforge.net/lists/listinfo/fofredux-devel
>