----- Original Message -----
From: "Enrico Forestieri" <[EMAIL PROTECTED]>
To: <lyx-users@lists.lyx.org>
Sent: Tuesday, December 13, 2005 7:02 PM
Subject: Re: There's Something About Textclass.lst [WinXP, installing into]
Stephen Harris <[EMAIL PROTECTED]> writes:
I wish I were smarter so that it would be blindingly obvious to me too.
I envision four(5) scenarios:
I see that Angus already answered your questions.
I think that your doubts are due to a misconception of the PATH.
Please, have a look here http://www.robvanderwoude.com/which.html
where you can find a script that scans the PATH to find a command.
--
Enrico
No, he did not. I'll get back to that. I appreciate your calm response.
This is what Angus initially said to me:
"The installer searches the Registry in various ways in order to find the
whereabouts of sh.exe, gswinc.exe, python.exe, etc. It's blindingly
obvious that it should first ascertain whether these things are already in
the PATH. Never mind ;-)"
SH: I did not say that the LyX installer could not search the PATH.
I said that this didn't contribute any information that couldn't be
acquired by other means. Since the PATH can contain duplicated
and non-existent entities it can't confirm reports from other means.
I've just installed LyX without LyX or any helper applications in
the Windows PATH environment variable. And it works. So can
you tell me why it is "blindingly obvious" that LyX should first
ascertain that these things are already in the PATH?
It is not that I disagree with his explanation; it is that his explanation
is not consistent with his initial claim, nor does it support that claim.
Angus: "Formally, that's the PATH environment variable. To be
honest, I don't think that I check the contents of this variable when
generating the contents of \path_prefix. Clearly I should have ;-)
SH: LyX1.3.6 works and it doesn't check/ascertain the PATH
variable and it was released prior to 1.3.7. So not checking the
PATH first is not the cause of the problem; is it supposed to
be "blindingly obvious" that checking the PATH should have
been used as a safety net? Even so, finding the executables
happens much earlier than outputting a too large \path_prefix.
So this part is not blindingly obvious to me because I don't see
it as following from his explanation, which seems reasonable.
------------------------------------------------------------------------------
And does checking the prepended LyX discovered executable
paths have to involve a comparison of the Windows Path
environment variable, or, perhaps some other variable could
store the information for later comparison with configure and
its output to \path_prefix?
This comment is speculative on my part, and perhaps a good
programmer could instantly dismiss this notion and say that
it is blindingly obvious that the Win PATH should be employed
for this. I am really not going to defend this last speculation as
refuting any claim of blindingly obvious. It is my comments
above the dotted line that I think refute claims of a blindingly
obvious situation. I wanted to make that clear because of the
tendency of people to commit the logical fallacy of blowing
up a minor claim/speculation and responding to that instead
of the heart of the position. I am disputing the "blindingly
obvious" statement, not the later explanation which in this
particular case I largely already knew, unlike for instance Aspell.
When I earned my computer degree I received two A's in Linux
classes which means I am know about the 'which' command,
although my programming skills are comparitively weak. Now
I have two winxp pro computers, one of which installs 137pre5
and the other one doesn't, and they both have similar versions
of helper applications installed.
I find that perplexing and a nice troubleshooting challenge.
Regards,
Stephen
