http://stlab.adobe.com/wiki/index.php?title=Argument_Against_GC&feed=atom&action=historyArgument Against GC - Revision history2013-05-23T15:16:39ZRevision history for this page on the wikiMediaWiki 1.19.0http://stlab.adobe.com/wiki/index.php?title=Argument_Against_GC&diff=2046&oldid=prevSeanParent: /* Conclusions */2007-09-24T22:30:04Z<p><span dir="auto"><span class="autocomment">Conclusions</span></span></p>
<table class='diff diff-contentalign-left'>
<col class='diff-marker' />
<col class='diff-content' />
<col class='diff-marker' />
<col class='diff-content' />
<tr valign='top'>
<td colspan='2' style="background-color: white; color:black;">← Older revision</td>
<td colspan='2' style="background-color: white; color:black;">Revision as of 22:30, 24 September 2007</td>
</tr><tr><td colspan="2" class="diff-lineno">Line 23:</td>
<td colspan="2" class="diff-lineno">Line 23:</td></tr>
<tr><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>== Conclusions ==</div></td><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>== Conclusions ==</div></td></tr>
<tr><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>* Complex solution to problem for which simple solutions are widely used and will be improved by C++0x leaving us little need.</div></td><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>* Complex solution to problem for which simple solutions are widely used and will be improved by C++0x leaving us little need.</div></td></tr>
<tr><td colspan="2"> </td><td class='diff-marker'>+</td><td style="background: #cfc; color:black; font-size: smaller;"><div><ins style="color: red; font-weight: bold; text-decoration: none;">* We have little to no experience with the recommended language features which are to be standardized.</ins></div></td></tr>
<tr><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>* Fixing bad software complex system will never work.</div></td><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>* Fixing bad software complex system will never work.</div></td></tr>
<tr><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>* Recommend minor language changes to improve future GC support - disallow hiding of pointers (xor list trick) as one example.</div></td><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>* Recommend minor language changes to improve future GC support - disallow hiding of pointers (xor list trick) as one example.</div></td></tr>
<tr><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"></td><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"></td></tr>
<tr><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>* Finally - address the "C++ is bad because it has no GC" argument head-on. C++ doesn't generate garbage and so has no need for GC. Clearly Java, C#, Objective C, etc. generate lots of garbage.</div></td><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>* Finally - address the "C++ is bad because it has no GC" argument head-on. C++ doesn't generate garbage and so has no need for GC. Clearly Java, C#, Objective C, etc. generate lots of garbage.</div></td></tr>
</table>SeanParenthttp://stlab.adobe.com/wiki/index.php?title=Argument_Against_GC&diff=2045&oldid=prevSeanParent at 22:27, 24 September 20072007-09-24T22:27:49Z<p></p>
<table class='diff diff-contentalign-left'>
<col class='diff-marker' />
<col class='diff-content' />
<col class='diff-marker' />
<col class='diff-content' />
<tr valign='top'>
<td colspan='2' style="background-color: white; color:black;">← Older revision</td>
<td colspan='2' style="background-color: white; color:black;">Revision as of 22:27, 24 September 2007</td>
</tr><tr><td colspan="2" class="diff-lineno">Line 8:</td>
<td colspan="2" class="diff-lineno">Line 8:</td></tr>
<tr><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>* For trash collection - sight issues with trash and reference counted systems [I had a good one recently - can't find it at the moment - bug easy to google for.]</div></td><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>* For trash collection - sight issues with trash and reference counted systems [I had a good one recently - can't find it at the moment - bug easy to google for.]</div></td></tr>
<tr><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>== Complexity of Current GC for C++ Design ==</div></td><td class='diff-marker'> </td><td style="background: #eee; color:black; font-size: smaller;"><div>== Complexity of Current GC for C++ Design ==</div></td></tr>
<tr><td class='diff-marker'>−</td><td style="background: #ffa; color:black; font-size: smaller;"><div>* Type qualifiers vs. modifiers for gc_strict and gc_relaxed</div></td><td class='diff-marker'>+</td><td style="background: #cfc; color:black; font-size: smaller;"><div>* Type qualifiers vs. modifiers <ins class="diffchange diffchange-inline">vs. attribute </ins>for gc_strict and gc_relaxed <ins class="diffchange diffchange-inline">- are none of these. We have no experience with such constructs and how they interact with type system.</ins></div></td></tr>
<tr><td colspan="2"> </td><td class='diff-marker'>+</td><td style="background: #cfc; color:black; font-size: smaller;"><div><ins class="diffchange diffchange-inline">* The new operator has some unspecified magic which allows it to communicate with the type system to know about strict/weak declarations. This information is not available to anyone writing a replacement allocator (which is quite common) and makes it impossible to use an existing application GC allocator (such as sharing an allocator with JavaScript or Managed C++) unless the system provides it.</ins></div></td></tr>
<tr><td colspan="2"> </td><td class='diff-marker'>+</td><td style="background: #cfc; color:black; font-size: smaller;"><div><ins class="diffchange diffchange-inline">* The proposal breaks conventions for RAII - and we have no experience or guidelines. When should one call delete?</ins></div></td></tr>
<tr><td colspan="2"> </td><td class='diff-marker'>+</td><td style="background: #cfc; color:black; font-size: smaller;"><div><ins class="diffchange diffchange-inline">* The gc_required and gc_forbidden keywords bifurcate the language.</ins></div></td></tr>
<tr><td colspan="2"> </td><td class='diff-marker'>+</td><td style="background: #cfc; color:black; font-size: smaller;"><div><ins class="diffchange diffchange-inline">* Suppress and collect must be used for critical areas of performance.</ins></div></td></tr>
<tr><td colspan="2"> </td><td class='diff-marker'>+</td><td style="background: #cfc; color:black; font-size: smaller;"><div><ins class="diffchange diffchange-inline">* No guidance on how multiple collectors in the same application will interact.</ins></div></td></tr>
<tr><td colspan="2"> </td><td class='diff-marker'>+</td><td style="background: #cfc; color:black; font-size: smaller;"><div> </div></td></tr>
<tr><td colspan="2"> </td><td class='diff-marker'>+</td><td style="background: #cfc; color:black; font-size: smaller;"><div><ins class="diffchange diffchange-inline">* Note that the vast amount of complexity necessary for GC is apparent in the number of keywords and library routines to support it. This is a strong argument that it is of little use for saving bad application - it is just as likely that GC will be misused (if not more so) then memory. Google "unmanaged pointers managed C++" for many examples. Note that as long as we support the C libraries (and C runtime) such issues will always be part of C++.</ins></div></td></tr>
<tr><td colspan="2"> </td><td class='diff-marker'>+</td><td style="background: #cfc; color:black; font-size: smaller;"><div><ins class="diffchange diffchange-inline">== Outline a Proposal to Address Some Issues ==</ins></div></td></tr>
<tr><td colspan="2"> </td><td class='diff-marker'>+</td><td style="background: #cfc; color:black; font-size: smaller;"><div><ins class="diffchange diffchange-inline">* Eliminate gc_required</ins></div></td></tr>
<tr><td colspan="2"> </td><td class='diff-marker'>+</td><td style="background: #cfc; color:black; font-size: smaller;"><div><ins class="diffchange diffchange-inline">* gc_strict as a type qualifier or attribute</ins></div></td></tr>
<tr><td colspan="2"> </td><td class='diff-marker'>+</td><td style="background: #cfc; color:black; font-size: smaller;"><div><ins class="diffchange diffchange-inline">* Definition of a operator new() in the presence of GC</ins></div></td></tr>
<tr><td colspan="2"> </td><td class='diff-marker'>+</td><td style="background: #cfc; color:black; font-size: smaller;"><div><ins class="diffchange diffchange-inline">* Use of gc_strict as part of type to guide and check RAII.</ins></div></td></tr>
<tr><td colspan="2"> </td><td class='diff-marker'>+</td><td style="background: #cfc; color:black; font-size: smaller;"><div><ins class="diffchange diffchange-inline">== Conclusions ==</ins></div></td></tr>
<tr><td colspan="2"> </td><td class='diff-marker'>+</td><td style="background: #cfc; color:black; font-size: smaller;"><div><ins class="diffchange diffchange-inline">* Complex solution to problem for which simple solutions are widely used and will be improved by C++0x leaving us little need.</ins></div></td></tr>
<tr><td colspan="2"> </td><td class='diff-marker'>+</td><td style="background: #cfc; color:black; font-size: smaller;"><div><ins class="diffchange diffchange-inline">* Fixing bad software complex system will never work.</ins></div></td></tr>
<tr><td colspan="2"> </td><td class='diff-marker'>+</td><td style="background: #cfc; color:black; font-size: smaller;"><div><ins class="diffchange diffchange-inline">* Recommend minor language changes to improve future GC support - disallow hiding of pointers (xor list trick) as one example.</ins></div></td></tr>
<tr><td colspan="2"> </td><td class='diff-marker'>+</td><td style="background: #cfc; color:black; font-size: smaller;"><div> </div></td></tr>
<tr><td colspan="2"> </td><td class='diff-marker'>+</td><td style="background: #cfc; color:black; font-size: smaller;"><div><ins class="diffchange diffchange-inline">* Finally - address the "C++ is bad because it has no GC" argument head-on. C++ doesn't generate garbage and so has no need for GC. Clearly Java, C#, Objective C, etc. generate lots of garbage.</ins></div></td></tr>
</table>SeanParenthttp://stlab.adobe.com/wiki/index.php?title=Argument_Against_GC&diff=2044&oldid=prevSeanParent at 22:07, 24 September 20072007-09-24T22:07:32Z<p></p>
<p><b>New page</b></p><div>== Use Cases for Garbage Collection in C++ ==<br />
* Object oriented programming - reference based objects<br />
* Functional programming - avoiding copies of immutable objects<br />
* Trash collection - preventing bad code from crashing or exhausting memory or address space.<br />
== Analysis of Each Use Case ==<br />
* Current solutions - reference counting and pooled memory<br />
* Look at what happens with reference counting with move semantics<br />
* For trash collection - sight issues with trash and reference counted systems [I had a good one recently - can't find it at the moment - bug easy to google for.]<br />
== Complexity of Current GC for C++ Design ==<br />
* Type qualifiers vs. modifiers for gc_strict and gc_relaxed</div>SeanParent