Page 31 of 33
Re: New designs for TPFC by tproli
Posted: Tue Mar 24, 2015 2:18 pm
by tproli
In the Add entry section, this label points to a wrong input:
Code: Select all
<label for="x_topic_id">Similar/alternative apps</label>
It should be
Code: Select all
<label for="x_alternatives">Similar/alternative apps</label>
Re: New designs for TPFC by tproli
Posted: Wed Mar 25, 2015 7:19 pm
by Andrew Lee
Corrected. Thanks!!
Re: New designs for TPFC by tproli
Posted: Thu Mar 26, 2015 12:15 am
by tproli
Great, thanks.
Re: New designs for TPFC by tproli
Posted: Wed May 06, 2015 5:01 am
by Craunch
There is a link from a program's entry in the collection to its forum topic, but there is no corresponding link from a program's forum topic to its software collection entry unless one of the forum entries happens to have included the link. Would it be useful to add such a link?
Re: New designs for TPFC by tproli
Posted: Wed May 06, 2015 5:14 am
by Midas
Sure. I always try to include such a link in my relevant posts...
Re: New designs for TPFC by tproli
Posted: Sun Aug 30, 2015 9:42 am
by joby_toss
Hi!
I didn't change any settings lately (
Newer first option is still active), yet I see this:
Re: New designs for TPFC by tproli
Posted: Sun Aug 30, 2015 10:12 am
by Checker
@joby_toss: I noticed the same. A lot of "old" stuff came up to the main page
Re: New designs for TPFC by tproli
Posted: Sun Aug 30, 2015 10:15 am
by SYSTEM
joby_toss wrote:Hi!
I didn't change any settings lately (
Newer first option is still active), yet I see this:
I think it happened because a ton of entries became public at once. Looks like DarkShadow and TP109 achieved rank 5, which brought a lot of entries to the ten point threshold.
The site orders entries by the time they have been updated or they have become public (whichever is later), and those entries became public
at the same time.
Re: New designs for TPFC by tproli
Posted: Sun Aug 30, 2015 10:36 am
by Checker
@SYSTEM: That's what I was thinking, too
But: ExperienceIndexOK (added/voted by me "R5" and voted by DarkShadow "R5") is public for a long time now (2015-07-06) and was updated by me several times (flagged "updated") now it appears as "new"
Same for "BowPad".
Added by DarkShadow 2015-01-20, public and updated by me several times (Jan, Feb, Mar, Apr, Jul) this year, is now again flagged as "new"
Re: New designs for TPFC by tproli
Posted: Sun Aug 30, 2015 10:48 am
by Checker
@SYSTEM: BTW: DarkShadow voted as R5 for some of my entries already.
Another example: "FlicFlac"
Even if DarkShadow and TP109 were not R5 till today, a vote of 13 points (R5+R4+R4) would be enough to be public.
And: If it's a new public entry it has to be flagged as "new" ... what it isn't
Re: New designs for TPFC by tproli
Posted: Sun Aug 30, 2015 6:15 pm
by Andrew Lee
I am looking into this problem. At first glance, it appears to be a code error, but I can't be sure yet.
It's rather tricky and will probably take some time to figure out.
Will keep you guys updated.
Re: New designs for TPFC by tproli
Posted: Sun Aug 30, 2015 7:28 pm
by Andrew Lee
What I've done so far: I have restored the "display_order" and "status" fields of the affected entries. A total of 78 entries were affected. This is easily done because the affected entries had the same display_order value.
I have to admit after many hours of tracing, I am still not quite sure what happened. The code seems alright. Here's the low-level details. The 3 fields you should care about are "updated", "display_order" and "status". The "updated" field changes every time an entry is modified. The "display_order" field only change when an entry is modified and the rules deem that it should move to the top of the list. The "status" field indicates whether an entry is private, new or updated.
In this case for all 78 entries, the "updated" field were untouched. The "display_order" field were changed to the same value. The "status" field were changed to "new". The scores remained the same i.e. no new votes were received during this period.
The only place where the "display_order" field is changed without updating the "updated" field is in the maintenance task. However, the code here is very simple. It will only update the "display_order" field for entries that are private. Yet for all 78 entries, their statuses were either "New" or "Updated" before the glitch hits.
So I have performed the above data restoration and will watch carefully if the glitch hits again. If it does, hopefully we'll have more clues to figure out what happened.
Re: New designs for TPFC by tproli
Posted: Sun Aug 30, 2015 11:02 pm
by SYSTEM
Andrew Lee wrote:What I've done so far: I have restored the "display_order" and "status" fields of the affected entries. A total of 78 entries were affected. This is easily done because the affected entries had the same display_order value.
I have to admit after many hours of tracing, I am still not quite sure what happened. The code seems alright. Here's the low-level details. The 3 fields you should care about are "updated", "display_order" and "status". The "updated" field changes every time an entry is modified. The "display_order" field only change when an entry is modified and the rules deem that it should move to the top of the list. The "status" field indicates whether an entry is private, new or updated.
In this case for all 78 entries, the "updated" field were untouched. The "display_order" field were changed to the same value. The "status" field were changed to "new". The scores remained the same i.e. no new votes were received during this period.
The only place where the "display_order" field is changed without updating the "updated" field is in the maintenance task. However, the code here is very simple. It will only update the "display_order" field for entries that are private. Yet for all 78 entries, their statuses were either "New" or "Updated" before the glitch hits.
So I have performed the above data restoration and will watch carefully if the glitch hits again. If it does, hopefully we'll have more clues to figure out what happened.
Thanks for the details.
I wonder if this was another SQL injection attack...
Re: New designs for TPFC by tproli
Posted: Mon Aug 31, 2015 3:05 am
by Andrew Lee
I wonder if this was another SQL injection attack...
Won't say it is impossible, but two arguments against this possibility.
1) All SQL calls are sanitized since the last SQL injection attack a long time back
2) There's no common pattern behind the entries affected. Kind of strange to formulate an SQL injection attack like this:
Code: Select all
update software set display_order=now(), status=1 where id in (id1, id2, id3, ...., id78)
All for what?
Re: New designs for TPFC by tproli
Posted: Mon Aug 31, 2015 6:08 am
by Midas
SYSTEM wrote:I think it happened because a ton of entries became public at once. Looks like DarkShadow and TP109 achieved rank 5, which brought a lot of entries to the ten point threshold.
The site orders entries by the time they have been updated or they have become public (whichever is later), and those entries became public at the same time.
Andrew Lee wrote:What I've done so far: [...]
- Judging from the newsfeed alone, that update was almost in the hundreds. I was trying to figure what had happened on my own, but just didn't manage. So from myself (and on behalf of any of our more dimwitted members) thank you ever so much for explaining it.
