[TYPO3-core] RFC #12231: New caching framework (4.3-dev) does not work with DBAL (oracle) [nobrainer]

[Xavier Perseguers]

Hi,

This RFC is deprecated in favor of RFC #12440.

Thanks for your understanding.



Xavier Perseguers wrote:
> Hi,
> 
> This is an SVN patch request.
> 
> Type: Bugfix / nobrainer (!)
> 
> Bugtracker reference:
> http://bugs.typo3.org/view.php?id=12231
> 
> Branches: trunk
> 
> Problem:
> 
> Do not go away! Solution is very easy :-)
> 
> When looking in the DBAL debug module, a lot of errors are produced, 
> which all point to the new caching framework. This error message is 
> "ORA-00904: "LIFETIME": invalid identifier".
> 
> This error is produced by 
> t3lib/cache/backend/class.t3lib_cache_backend_dbbackend, on line 96:
> . 'AND (crdate + lifetime >= ' . time() . ' OR lifetime = 0)'
> and probably in the same file on line 120:
> . 'AND crdate + lifetime >= ' . time()
> 
> Two fields are used in the addition; crdate + lifetime. Lifetime is not 
> recognized as a fieldname and therefore is not quoted with double quotes 
> (")
> 
> The query ends up like:
> 
> SELECT "content"
> FROM "cache_hash"
> WHERE
> "identifier" = 'abbbabaf2d4b3f9a63e8dde781f1c106' AND 
> ("crdate"+lifetime) >= 1255507193 OR "lifetime" = 0)
> 
> which should be:
> 
> SELECT "content"
> FROM "cache_hash"
> WHERE
> "identifier" = 'abbbabaf2d4b3f9a63e8dde781f1c106' AND 
> ("crdate"+"lifetime" >= 1255507193 OR "lifetime" = 0)
> 
> After investigation, I found that the problem relies mainly in Core SQL 
> parser that is not able to detect that the "calc_value" (naming from 
> parser) is a field name and as such there is no chance for DBAL to 
> detect this and quote it/remap it accordingly.
> 
> Solution:
> Add code in t3lib_sqlparser to detect that a field was encountered and 
> store its information (possible table.field combination) just as the 
> first part of the calc clause was stored, with a table containing the 
> field and the table name. As remapping is only used by DBAL, do not 
> modify existing key (=> no change when not using DBAL = _nobrainer_ and 
> no change for those of us using DBAL until I commit associated patch to 
> DBAL, which I'll do as soon as this one is in).
> 
> Thanks for taking care of getting this in. This is really a no-brainer!
> 


-- 
Xavier Perseguers
http://xavier.perseguers.ch/en

One contribution a day keeps the fork away