<?xml version="1.0" encoding="utf-8"?>
<feed version="0.3" xmlns="http://purl.org/atom/ns#" xmlns:dc="http://purl.org/dc/elements/1.1/" xml:lang="en">
<title>It&apos;s My Machine!</title>
<link rel="alternate" type="text/html" href="http://my-machine.mumble.net/" />
<modified>2005-04-11T02:35:12Z</modified>
<tagline>Alan Ruttenberg&apos;s outlet for lisp/OpenMCL and Mac stuff</tagline>
<id>tag:my-machine.mumble.net,2008://1</id>
<generator url="http://www.movabletype.org/" version="3.01D">Movable Type</generator>
<copyright>Copyright (c) 2005, alanr</copyright>
<entry>
<title>Large address space SBCL on OS X</title>
<link rel="alternate" type="text/html" href="http://my-machine.mumble.net/archives/2005/04/large_address_s.html" />
<modified>2005-04-11T02:35:12Z</modified>
<issued>2005-04-11T02:09:13Z</issued>
<id>tag:my-machine.mumble.net,2005://1.14</id>
<created>2005-04-11T02:09:13Z</created>
<summary type="text/plain">I usually use OpenMCL for my development. I want to the code on x86 linux so I was testing it on in CMUCL on OS X, as a first step. To my amazement when timing a particularly long computation I...</summary>
<author>
<name>alanr</name>

<email>alanr-blog@mumble.net</email>
</author>
<dc:subject>Lisp</dc:subject>
<content type="text/html" mode="escaped" xml:lang="en" xml:base="http://my-machine.mumble.net/">
<![CDATA[<p>I usually use <a href="http://openmcl.clozure.com/">OpenMCL</a> for my development. I want to the code on x86 linux so I was testing it on in <a href="http://www.cons.org/cmucl/">CMUCL</a> on OS X, as a first step. To my amazement when timing a particularly long computation I found that CMUCL ran 6x faster than my openMCL version. When I attempted to port a second computation I ran out of address space. Stock CMUCL runs with (I can't remember at the moment) either 128M or 256M. Neither was enough. I didn't immediately see what to do about that so I started looking at <a href="http://www.sbcl.org/">SBCL</a>. With some help from <a href="http://www.cs.indiana.edu/~bmastenb/">Brian Mastenbrook</a> I adjusted <a href="http://cvs.sourceforge.net/viewcvs.py/sbcl/sbcl/src/compiler/ppc/parms.lisp?view=markup">src/compiler/ppc/parms.lisp</a> and recompiled using openmcl as the cross compiler. The parameters I changed were: </p>

<pre>
(def!constant dynamic-space-start   #x10000000)
(def!constant dynamic-space-end     #x3FFFF000)
(def!constant dynamic-0-space-start #x10000000)
(def!constant dynamic-0-space-end   #x3Ffff000)
(def!constant dynamic-1-space-start #x40000000)
(def!constant dynamic-1-space-end   #x6ffff000)
(def!constant linkage-table-space-start #x70000000)
(def!constant linkage-table-space-end   #x71000000)
</pre>

<p>This gives me about 800M of address space which should serve my needs.  Note that SBCL currently does a stop and copy GC, which means you could need 1.6G at peak.</p>

<p>Another gotcha is that I was getting errors when building and running sbcl: <i>"*** malloc[2292]: Deallocation of a pointer not malloced: 0x7800400; This could be a double free(), or free() called with the middle of an allocated block; Try setting environment variable MallocHelp to see tools to help debug"</i>. Brian asked me what my setting for DYLD_LIBRARY_PATH was. On my machine it is /Developer/qt/lib. Removing that caused my problems to go away.</p>]]>

</content>
</entry>
<entry>
<title>Etymotic ER-6i earbuds for iPod</title>
<link rel="alternate" type="text/html" href="http://my-machine.mumble.net/archives/2005/04/etymotic_er6i_e.html" />
<modified>2005-04-11T02:07:20Z</modified>
<issued>2005-04-11T01:40:12Z</issued>
<id>tag:my-machine.mumble.net,2005://1.13</id>
<created>2005-04-11T01:40:12Z</created>
<summary type="text/plain">Just got these last week and they are amazing! As much as getting the iPod was an amazing improvement to my life, so these are on top of that. What&apos;s good about them They isolate outside noise very well. I&apos;m...</summary>
<author>
<name>alanr</name>

<email>alanr-blog@mumble.net</email>
</author>
<dc:subject>Misc</dc:subject>
<content type="text/html" mode="escaped" xml:lang="en" xml:base="http://my-machine.mumble.net/">
<![CDATA[<p>Just got <a href="http://www.etymotic.com/ephp/er6i.asp">these</a> last week and they are amazing! As much as getting the iPod was an amazing improvement to my life, so these are on top of that. </p>

<p><b>What's good about them</b></p>

<p>They isolate outside noise very well. I'm already a heavy user of foam earplugs as I find I am inordinately sensitive to noise in a number of contexts. I use <a href="http://www.labsafety.com/store/product_group.asp?dept_id=12619&parent_id=10430">Howard Leight Max Earplugs</a>, in case you are looking for some pretty effective ones.  The foam tips for the ER-6i are smaller than the usual foam earplugs I use, and more comfortable. But they feel comparable in the amount of noise they block out. For comparison, the Leights are rated at 33db reduction and the ER-6i are rated at 27db. I walk for exercise and when using the Apple earbuds it was very annoying because, when listening to a radio show or a talk, if  a car went by it would prevent me from hearing what was said. Not a problem any more. </p>

<p>The sound is amazing. I keep hearing things that I've not hear before in music I've listed to a lot. With the Apple earbuds I would change the EQ settings on my ipod and wondered whether there was any difference. When using the ER-6i there is a clear difference in EQ settings. The sound is so pleasurable that it's got me listening to a lot more music for the sheer sensual pleasure of it!</p>

<p><b>What could be improved</b></p>

<p>You hear any jostling of the wires. I have to wonder, since Etymotic has such obviously talented sound engineers, whether it was a cost cutting effort that prevented them from including some better coupling of the wires to the earbuds to reduce this. The clip that comes with is helpful - if you are wearing something to clip it to. But this will be a problem when walking in the hot summer.</p>

<p>Related to the transmission of sound from the wires to the earbuds, when the wind blows, and not too much wind at that, you can hear a whistling sound transmitted by the wires. An unfortunate resonance.  You can avoid this somewhat by arranging the wires so that they hang closer to your head  - I drape them over my ears and this helps somewhat.</p>]]>

</content>
</entry>
<entry>
<title>SafariStand</title>
<link rel="alternate" type="text/html" href="http://my-machine.mumble.net/archives/2005/04/safaristand.html" />
<modified>2005-04-08T18:21:24Z</modified>
<issued>2005-04-08T17:44:48Z</issued>
<id>tag:my-machine.mumble.net,2005://1.12</id>
<created>2005-04-08T17:44:48Z</created>
<summary type="text/plain">SafariStand is a Safari plugin that offers a number of useful extensions. The ones I find useful are Show Page Info which lets me see the http headers for a page Related Cookies which lets me look at the cookies...</summary>
<author>
<name>alanr</name>

<email>alanr-blog@mumble.net</email>
</author>
<dc:subject>Web Browsing</dc:subject>
<content type="text/html" mode="escaped" xml:lang="en" xml:base="http://my-machine.mumble.net/">
<![CDATA[<p><a href="http://hetima.com/safari/stand-e.html">SafariStand</a> is a Safari plugin that offers a number of useful extensions. The ones I find useful are <br />
<ol><li><b>Show Page Info</b> which lets me see the http headers for a page<br />
<li><b>Related Cookies</b> which lets me look at the cookies that a page is using. This was particularly helpful when I had to do some screen scraping off a web page that had a session id. Previously I was using <a href="http://www.baurhome.net/software/index.html">Eavesdrop</a> to sniff the traffic for a page and grab the cookie from the transcript. This is faster, since it is available from a menu inside the browser.<br />
<li><b>Quick Search</b> This lets you select some text from a page and have the context menu come up offering to do a search using those words on the search engine of your choice. I have it set up to use<br />
<ul><li>Google: http://www.google.com/search?q=@key&amp;ie=UTF-8&amp;oe=UTF-8<br />
<li>Google Scholar: http://scholar.google.com/search?q=@key&amp;ie=UTF-8&amp;oe=UTF-8<br />
<li>Acronyms: http://it.acronymfinder.com/af-query.asp?string=exact&amp;acronym=@key<br />
<li>Dictionary: http://dictionary.reference.com/search?q=@key<br />
<li>Wikipedia: http://www.google.com/search?hl=en&amp;q=wikipedia+@key&amp;btnI=I%27m+Feeling+Lucky<br />
</ul><li><b>Editable source view</b> makes the <i>view source</i> window editable. You can now change the html under a page and have the browser redisplay the page with your changes. Handy for debugging web applications, for fixing broken pages that you need, and for removing the occasional AUTOCOMPLETE=off directive from a page so that Safari will remember your login information<br />
</ol></p>

<p>The documentation wasn't clear on how to activate the menu items for <i>Show Page Info</i>. The way it works is that you tell SafariStand which browser button you want it to take over. Then make sure that that button is visible, i.e. checked off in Safari's <i>View</i> menu. I used the <i>Bug</i> button. SafariStand changes the icon of that button and shows its menu when you click the button.</p>]]>

</content>
</entry>
<entry>
<title>Slow Address Book</title>
<link rel="alternate" type="text/html" href="http://my-machine.mumble.net/archives/2005/04/slow_address_bo_1.html" />
<modified>2005-04-08T17:43:42Z</modified>
<issued>2005-04-08T17:37:21Z</issued>
<id>tag:my-machine.mumble.net,2005://1.11</id>
<created>2005-04-08T17:37:21Z</created>
<summary type="text/plain">My OS X Address Book application was glacially slow - it might take 20 seconds between when I clicked the checkbox indicating that the entry was a company, to when it responded. Looked at the Address Book data file: ~/Library/Application...</summary>
<author>
<name>alanr</name>

<email>alanr-blog@mumble.net</email>
</author>
<dc:subject>OS X</dc:subject>
<content type="text/html" mode="escaped" xml:lang="en" xml:base="http://my-machine.mumble.net/">
<![CDATA[<p>My OS X Address Book application was glacially slow - it might take 20 seconds between when I clicked the checkbox indicating that the entry was a company, to when it responded.</p>

<p>Looked at the Address Book data file: ~/Library/Application Support/AddressBook/AddressBook.data and saw that it was 100M! Looking inside, I realized that it contained the the previous recipients list from Mail.app, and that this was the bulk of it's contents - presumably from all the addresses of spam I received. Following the hint on macosxhints <a href="http://www.macosxhints.com/article.php?story=2004102302295562&query=previous+recipients">Display previous Mail.app recipient information</a> I found that I could delete previous recipients. I did that, and the file shrunk to 3M!</p>

<p>Address Book runs substantially faster now.</p>]]>

</content>
</entry>
<entry>
<title>Acrobat javascript annoyance</title>
<link rel="alternate" type="text/html" href="http://my-machine.mumble.net/archives/2005/03/acrobat_javascr.html" />
<modified>2005-03-31T18:32:20Z</modified>
<issued>2005-03-31T18:20:20Z</issued>
<id>tag:my-machine.mumble.net,2005://1.10</id>
<created>2005-03-31T18:20:20Z</created>
<summary type="text/plain">You might want to disable javascript in Adobe Acrobat 7, either because you are generally paranoid, like me, or specifically paranoid, like Paul Amoroso. The problem is that when I did this, Acrobat asked me, every time I quit it...</summary>
<author>
<name>alanr</name>

<email>alanr-blog@mumble.net</email>
</author>
<dc:subject>Misc</dc:subject>
<content type="text/html" mode="escaped" xml:lang="en" xml:base="http://my-machine.mumble.net/">
<![CDATA[<p>You might want to disable javascript in Adobe Acrobat 7, either because you are generally paranoid, like me, or specifically paranoid, like <a href="http://www.paoloamoroso.it/log/050331.html">Paul Amoroso</a>.</p>

<p>The problem is that when I did this, Acrobat asked me, every time I quit it or Safari, which uses the PDF plugin, whether it could turn it back on because some file might not work without it. Here's how I fixed that.</p>

<p>Using lsof when the annoying dialog was up, I found that the file in question is <i>~/Library/Acrobat User Data/7.0/JavaScripts/glob.settings.js</i>. I edited the file and removed the contents, which was essentially an empty list. After finding that Acrobat recreates the file contents, I locked the file and changed the permissions to read only. That does the trick. Now I don't get the complaint.</p>]]>

</content>
</entry>
<entry>
<title>OpenMCL: Persisting xref information in fasl files</title>
<link rel="alternate" type="text/html" href="http://my-machine.mumble.net/archives/2004/09/persisting_xref.html" />
<modified>2004-09-28T06:13:07Z</modified>
<issued>2004-09-28T05:41:14Z</issued>
<id>tag:my-machine.mumble.net,2004://1.8</id>
<created>2004-09-28T05:41:14Z</created>
<summary type="text/plain">Oliver Markovic added a xref implementation to OpenMCL last year, enabling one to look up callers of function, references to variables, etc. His implementation does not save information in compiled files. So if you compile your files and load them...</summary>
<author>
<name>alanr</name>

<email>alanr-blog@mumble.net</email>
</author>
<dc:subject>Lisp</dc:subject>
<content type="text/html" mode="escaped" xml:lang="en" xml:base="http://my-machine.mumble.net/">
<![CDATA[<p><a href="http://www.cliki.net/Oliver%20Markovic">Oliver Markovic</a> added a <a href="http://clozure.com/pipermail/openmcl-devel/2003-December/001253.html">xref implementation to OpenMCL</a> last year, enabling one to look up callers of function, references to variables, etc. His implementation does not save information in compiled files. So if you compile your files and load them into a fresh world then you lose the xref information. This file <a href="/files/xref-fasl.lisp">xref-fasl.lisp</a> implements that. I also added support for the xref implementation to <a href="http://www.common-lisp.net/project/slime/">SLIME</a>. Try slime-who-calls and friends from the version of slime in cvs.</p>]]>

</content>
</entry>

</feed>
