On Tue, Mar 26, 2013 at 4:35 PM, Fernando Cassia <fcas...@gmail.com> wrote: > On Tue, Mar 26, 2013 at 9:52 AM, Rob Weir <robw...@apache.org> wrote: > >> My ordered preference would be 1, 3 > > > yeah, great *sarcasm*, let's add another bullet point to a microsoft > presentation titled 'how Microsoft Office is better than > OpenOfffice.org".... OO.o lacks database? check!" > > My opinion is that maybe Sun put HSQLDB in there to fill in the need for a > resident database engine, which in the commercial offering (StarOffice) was > filled by Adabas D. > > One can still read the positive reviews of StarOffice where the database > module is praised: > > http://www.amazon.com/Sun-Microsystems-0614647643195-StarOffice-7/dp/B0000DG2N4 > > //// > * StarOffice Adabas (database application) is included (getting MS Access > requires buying MS Office Pro) and is easier to use than MS Access. Adabas > integrates with other StarOffice apps so, for instance, users can easily > create mail merge documents. > /// > > So, if HSQLDB is not up to par, maybe the realistic solution is to find a > database engine lightweight and powerful enough to take the role that > Adabas D had in StarOffice?. > > FC > PS: I read this whole thread as 'we don't want to maintain this code, since > we don't understand it, and we fear it's buggy'. But the solution in any > case is replacing the database module for another, or improving the > existing code, not making excuses for saying 'people don't use databases > after all so it should be gone'.
+1 Again I ask: where are the Bugzilla bugs? A quick search finds a corruption bug (https://issues.apache.org/ooo/show_bug.cgi?id=55496) which states: "While it's difficult to fix this with our current approach, where the data is just some sub-streams in a ZIP package, we need to somehow address this. I fear this might in fact only be possible with a package format (replacing the ZIP package format for this purpose) which allows O(1) random access to all of its parts." If that's the only problem, I don't think the author of that message realized that there is a way to get O(1) random access to the contents of a zipped file: ZIP files allow each archived file to use its own compression type, including no compression. HSQLDB could write directly to its uncompressed byte range in the ZIP file, with a thin shim checking for when it attempts to write beyond the end of this range (ie. grow the database), and then it could expand the ZIP file to accommodate it. With a bit of cleverness, this could even be done in-place, by keeping the HSQL database at the end of the ZIP file (when possible), and just moving up and updating the ZIP directories (and other files, when it's not at the end) that follow it. Oh and as for Jürgen's comment: +1 to competing with Microsoft, because any good office suite naturally will - and should. Damjan --------------------------------------------------------------------- To unsubscribe, e-mail: dev-unsubscr...@openoffice.apache.org For additional commands, e-mail: dev-h...@openoffice.apache.org
