Sign in to follow this  
Killocan

Software exorcism book.

Recommended Posts

Killocan    122
Hi, first of all, my apologizes to post in this place, but i can't post on (book) section, it's forbidden for me, so... I'm posting this issue to see what you guys think about :) I bought “Software exorcism – a handbook for debugging and optimizing legacy code” and a think I didn’t get this part: On page 207 the author states: "Some business software companies prefer to strip debug information and accept the associated performance hit, because it allows sales engineers to perform an on-site diagnosis. When sales engineers make a house call, all that they need to do in order to take a look under the hood is insert the debug information and crank up a debugger." That's sounds a little bit strange for me, because I can't see how removing debug information will hit performance down! Or if I get it wrong (performance will be improved) what's the problem?!?!! Performance gain is not a good thing? You know? :) PS: BAF(on allegro.cc) suggest-me that this statement is supposed to be written (not to strip debug...) I think that's right. Thanks, Marcos Emmanuel.

Share this post


Link to post
Share on other sites
theOcelot    498
That's probably a typo.

This site is weird about what it says you're allowed and not allowed to do. Are you sure you were logged in when you tried posting in the books section?

Share this post


Link to post
Share on other sites
Oluseyi    2103
Quote:
Original post by theOcelot
This site is weird about what it says you're allowed and not allowed to do. Are you sure you were logged in when you tried posting in the books section?

The books section isn't a regular forum; you can only post reviews or reply to reviews, you can't start arbitrary new threads.

Share this post


Link to post
Share on other sites
theOcelot    498
Ahhhh... It's still weird in other ways, though. It never can decide if I'm allowed to post a reply if I'm not logged in. Once it almost let me edit someone else's post.

Share this post


Link to post
Share on other sites
Oluseyi    2103
Quote:
Original post by theOcelot
Ahhhh... It's still weird in other ways, though. It never can decide if I'm allowed to post a reply if I'm not logged in. Once it almost let me edit someone else's post.

Not really. You can click edit and it will reveal the post and all its formatting. You can even click submit and the "You are about to edit someone else's post" dialog will pop up (that was really added for moderators), but if you clicked OK then you'd get a permission error.

It works as intended. It just doesn't communicate its intentions clearly. [smile]

Share this post


Link to post
Share on other sites
theOcelot    498
Figured it was a communication issue. But why does it only sometimes say that I can post replies when I'm not logged on? Sometimes the whole list at the bottom of the page is "may not x", while sometimes the "reply to thread" one is "you may post replies"? I hope I'm not hijacking this thread too badly, but that has always bugged me.

Share this post


Link to post
Share on other sites
Deyja    920
How I interpret what the book is saying is that they ship a debug build; taking the performance hit; but they strip the debug info.

So, all the on-site guy has to do is plug the debug info into the debug build and debug it.

Share this post


Link to post
Share on other sites
Killocan    122
Hummm... Deyja, i don't think so, this sounds a bit odd, because he (the author) don't say anything about it, look this is the "FULL" section:

One simple way to make debugging more expensive is to remove debugging information from your deliverable. This can be done by stripping debug information (with a tool like GNU's strip utility) or by setting your development tools to generate a release build.
Some business software companies prefer to strip debug information and accept the associated performance hit, because it allows sales engineers to perform an on-site diagnosis. When sales engineers make a house call, all that they need to do in order to take a look under the hood is insert the debug information and crank up a debugger.

...

Share this post


Link to post
Share on other sites
theOcelot    498
So they leave sort-of leave a placeholder for the debug info?
Quote:
One simple way to make debugging more expensive is...
What does that mean? Why is that a good thing???

Share this post


Link to post
Share on other sites
swiftcoder    18437
Quote:
Original post by Killocan
Hummm... Deyja, i don't think so, this sounds a bit odd, because he (the author) don't say anything about it, look this is the "FULL" section:

When sales engineers make a house call, all that they need to do in order to take a look under the hood is insert the debug information and crank up a debugger.
This is exactly what Deyja suggested. They ship a debug build, but they strip the debug info from it. Then all the programmer has to do is bring a copy of that stripped debug info, and he has a full debug build.

Share this post


Link to post
Share on other sites
Killocan    122
ok, but... look:

"This can be done by stripping debug information (with a tool like GNU's strip utility) or by setting your development tools to generate a release build."

He suggests to make a release build.

It's still not clear for me, all i can do is suppose. What Deyja said makes sense, but it's not explicity on the text.

Share this post


Link to post
Share on other sites
Deyja    920
Yes it is. They give you two options. Then proceed to say that many companies choose the former, which results in a performance hit compared to the latter.

So in short I was right.

Share this post


Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now

Sign in to follow this