PDA

View Full Version : Stale content in SC Cache



Philip Meyer
2007-11-22, 02:24
I am worried that content in the SC cache may contain stale data. A few times I have been asked to clear the cache to ensure strings are updated, etc.

Is this perhaps an effect of me running from SVN? I'd assume that on startup, if any of the timestamps for any strings file were newer than the cached strings.bin, then the strings.bin would be rebuilt, and all other cached content would have similar checks (just at startup?).

Phil

kdf
2007-11-22, 02:30
On 22-Nov-07, at 1:24 AM, Phil Meyer wrote:

> I am worried that content in the SC cache may contain stale data.
> A few times I have been asked to clear the cache to ensure strings
> are updated, etc.
>
> Is this perhaps an effect of me running from SVN? I'd assume that
> on startup, if any of the timestamps for any strings file were
> newer than the cached strings.bin, then the strings.bin would be
> rebuilt, and all other cached content would have similar checks
> (just at startup?).

Maybe, maybe not. but if what one user is running doesn't show the
same problems...we all start suggesting a clean slate.
Lets not lose focus trying to question the rationale for every little
thing. we cannot spend time walking everyone through every possible
thing to check.
If you want to avoid being asked to wipe things, best learn how to
check for yourself whether you have clean setup or not.

The startup will state whether strings are loaded from cache or not,
even stating the file location.
Please try to allocate some of the little free time that you have
familarising yourself with the log info, during startup and while
running.

As tiring as you feel it is being asked, it's just as tiring having
to ask.
-kdf

Philip Meyer
2007-11-22, 03:44
>If you want to avoid being asked to wipe things, best learn how to
>check for yourself whether you have clean setup or not.
>
In this particular case, I had already summised that the cache was not the issue, as I had cleaned it and restarted.

I can understand being asked to start from a clean slate when confirming other problems. I agree that this is a good thing to do. However, I was concerned that it should never be necessary to delete the cache. If support were to tell users to try flushing their cache and trying again, and this fixed issues, then we'd never get to find the reason for caches being stale.

I have a non-standard setup, so I was wondering not so much for my configuration, but how it may work for end-users. I don't personally mind having to flush the cache for example each time I perform an SVN update (as I could do it in an automated script), but wondered if being asked to flush the cache was because of known issues with the cache mechanism, server configuration, etc. I have started to delete the cache dir and browser cache and trying again before reporting any bugs.

>The startup will state whether strings are loaded from cache or not,
>even stating the file location.
>
I can confirm that for strings, it does appear to be working based on file timestamps, as if I touch a strings.txt file for a plugin, it does then report that it is loading from all source string files at the next startup.

Thanks.

Phil

kdf
2007-11-22, 03:56
On 22-Nov-07, at 2:44 AM, Phil Meyer wrote:

>
> I can understand being asked to start from a clean slate when
> confirming other problems. I agree that this is a good thing to
> do. However, I was concerned that it should never be necessary to
> delete the cache.
it shouldn't, but you aren't talking about a normal case. you are
talking about a bug comment that doesn't what happens for someone
else with a fresh svn checkout.

> If support were to tell users to try flushing their cache and
> trying again, and this fixed issues, then we'd never get to find
> the reason for caches being stale.
>
perhaps we need a "support" forum then, for discussion of support
training issues?


>> The startup will state whether strings are loaded from cache or not,
>> even stating the file location.
>>
> I can confirm that for strings, it does appear to be working based
> on file timestamps, as if I touch a strings.txt file for a plugin,
> it does then report that it is loading from all source string files
> at the next startup.

great, but until your buggered strings get sorted, you'll need to
learn what to look for or it begins to take more time than it's work,
given that the bug in question would appear to be fixed for 'the end
user'
-kdf

Philip Meyer
2007-11-22, 04:16
>it shouldn't, but you aren't talking about a normal case. you are
>talking about a bug comment that doesn't what happens for someone
>else with a fresh svn checkout.
>
I wasn't specifically talking about any specific bug. I have updated information about that specific issue in bugzilla.

>perhaps we need a "support" forum then, for discussion of support
>training issues?
>
Probably overkill.

>great, but until your buggered strings get sorted, you'll need to
>learn what to look for or it begins to take more time than it's work,
>given that the bug in question would appear to be fixed for 'the end
>user'
For this specific case, I don't think the strings are buggered, and I don't think it's a caching issue. I have a full fresh SVN checkout, and have cleared my cache. I can see the strings expanded correctly when trying to delete favorites from the favorites menu, but not from the screen after the menu where the favorite has been added from.

Phil