Comparison of string values provides wrong results for GE and LE operators. I provide example queries, notice the number of rows returned by the queries.
For query:

SELECT BQT1.SmallA.StringNum FROM BQT1.SmallA WHERE BQT1.SmallA.StringNum <= -22 ORDER BY StringNum

• Process Tree:
  Unable to find source-code formatter for language: plain. Available languages are: actionscript, ada, applescript, bash, c, c#, c++, cpp, css, erlang, go, groovy, haskell, html, java, javascript, js, json, lua, none, nyan, objc, perl, php, python, r, rainbow, ruby, scala, sh, sql, swift, visualbasic, xml, yaml

  LimitNode(0) output=[g_0.stringNum] limit 100
      AccessNode(1) output=[g_0.stringNum] SELECT g_0.stringNum FROM SmallAs.smallARemotecache AS g_0 WHERE g_0.stringNum <= '-22' ORDER BY g_0.stringNum
  

• SRC CMD:

  SELECT g_0.stringNum FROM SmallAs.smallARemotecache AS g_0 WHERE g_0.stringNum <= '-22' ORDER BY g_0.stringNum

• result 0 rows

But for query:

SELECT BQT1.SmallA.StringNum FROM BQT1.SmallA WHERE BQT1.SmallA.StringNum < -22 ORDER BY StringNum

• Process Tree:
  Unable to find source-code formatter for language: plain. Available languages are: actionscript, ada, applescript, bash, c, c#, c++, cpp, css, erlang, go, groovy, haskell, html, java, javascript, js, json, lua, none, nyan, objc, perl, php, python, r, rainbow, ruby, scala, sh, sql, swift, visualbasic, xml, yaml

  ProjectNode(0) output=[c.stringNum AS StringNum] [c.stringNum AS StringNum]
  	  LimitNode(1) output=[c.stringNum] limit 100
  	    SortNode(2) output=[c.stringNum] [SORT] [c.stringNum]
  	      SelectNode(3) output=[c.stringNum] c.stringNum < '-22'
  		AccessNode(4) output=[c.stringNum] SELECT g_0.stringNum FROM SmallAs.smallARemotecache AS g_0

• SRC CMD:

  SELECT g_0.stringNum FROM SmallAs.smallARemotecache AS g_0

• result 14 rows

And query:

SELECT BQT1.SmallA.StringNum FROM BQT1.SmallA WHERE BQT1.SmallA.StringNum = -22 ORDER BY StringNum

• Process Tree:
  Unable to find source-code formatter for language: plain. Available languages are: actionscript, ada, applescript, bash, c, c#, c++, cpp, css, erlang, go, groovy, haskell, html, java, javascript, js, json, lua, none, nyan, objc, perl, php, python, r, rainbow, ruby, scala, sh, sql, swift, visualbasic, xml, yaml

  LimitNode(0) output=[c.stringNum AS StringNum] limit 100
  		AccessNode(1) output=[c.stringNum AS StringNum] SELECT g_0.stringNum FROM SmallAs.smallARemotecache AS g_0 WHERE g_0.stringNum = '-22' ORDER BY g_0.stringNum

• SRC CMD:

  SELECT g_0.stringNum FROM SmallAs.smallARemotecache AS g_0 WHERE g_0.stringNum = '-22' ORDER BY g_0.stringNum

• result 1 row

The first query should then return 15 rows instead of 0. Also the queries differ in a way they are processed, the first one is pushed down to infinispan, the other two are processed by teiid, which is probably a regression originally tracked here: TEIID-3424

The same cause introduces problems with similar queries:

Select IntKey, StringKey From BQT1.SmallA WHERE NOT(StringKey > 10 AND IntKey < 47) ORDER BY IntKey

Which is processed as:

• Process Tree:
  Unable to find source-code formatter for language: plain. Available languages are: actionscript, ada, applescript, bash, c, c#, c++, cpp, css, erlang, go, groovy, haskell, html, java, javascript, js, json, lua, none, nyan, objc, perl, php, python, r, rainbow, ruby, scala, sh, sql, swift, visualbasic, xml, yaml

  LimitNode(0) output=[c.intKey AS IntKey, c.stringKey AS StringKey] limit 100
  		  AccessNode(1) output=[c.intKey AS IntKey, c.stringKey AS StringKey] SELECT g_0.intKey, g_0.stringKey FROM SmallAs.smallARemotecache AS g_0 WHERE (g_0.stringKey <= '10') OR (g_0.intKey >= 47) ORDER BY g_0.intKey