geekmaster's simple guide to tcc (tinycc)

[geekmaster]

You could probably force specific libraries to load first with -L, but there are already two known missing symlinks (and probably more), and the version numbers depend on which kindle firmware (tccmake works on DX/DXG/K3/K4/K5). Also, tccmake only supports script parameters $1 through $9, so for more parameters it would need to loop and shift. The more libraries you force, the less modules you can link, without adding complexity to the script.

When I searched for a solution to this problem (tcc only having PARTIAL support for the "ld scripting language"), the forums that discussed this problem called this a "library loader script" and I just continued to use that existing terminology. Because the solution that I used worked, I did not have to RTFM, so I did not learn the "right words" to call these things.

But even though tcc does not support all four ld scripting keywords, that is not the problem here. The real problem is that the old ld script points to a missing static link library, and tcc complains that it cannot find the library that the old "ld control script" pointed to. My simplified version removed the reference to the missing file (which may cause problems with static linking, but we are solving these problems as the need arises and that has not yet been a problem).

I just use tcc for testing simple stuff, and I use codesourcery for the final binary executable, which runs many time faster in most cases.

[knc1]

It was a simple "yes/no" question about your tcc not about your tccmake.

I think that the leading: "could probably" translates into "untested".
I can't read anything else in the noise that is relevant to my question past the first ", but ..."

But thank you for your reply.

[geekmaster]

Originally Posted by knc1

It was a simple "yes/no" question about your tcc not about your tccmake.

I think that the leading: "could probably" translates into "untested".
I can't read anything else in the noise that is relevant to my question past the first ", but ..."

But thank you for your reply.

Your way has complications. My previous post gives the details. If you want to know WHAT the complications are of using your suggested method, go back and read past the ", but" in my previous post. I did not say I did not do it. In fact I did try that (successfully) and then I rejected it as version-dependent. I decided to fix the missing symlink instead of forcing libraries (that may not even be used all the time). I want to keep my scripts small and simple.

[knc1]

Originally Posted by geekmaster

Your way has complications. My previous post gives the details. If you want to know WHAT the complications are of using your suggested method, go back and read past the ", but" in my previous post. I did not say I did not do it. In fact I did try that (successfully) and then I rejected it as version-dependent. I decided to fix the missing symlink instead of forcing libraries (that may not even be used all the time). I want to keep my scripts small and simple.

But why do you insist that everyone must make the same choices as yourself?
What are people here supposed to do that have objectives different than your own? Ignore you or feel free to ask what you have done and learned?

[geekmaster]

Originally Posted by knc1

But why do you insist that everyone must make the same choices as yourself?
What are people here supposed to do that have objectives different than your own? Ignore you or feel free to ask what you have done and learned?

I did not insist any such thing. People can add as much "noise" to their scripts and other code as they want to. The larger scripts do contain the MIT license, after all.

Why are you getting bent over my descriptions of how I *solved* the problems? And why argue over OTHER ways that these problems COULD have been solved? They ARE solved (until the next one comes along).

[geekmaster]

Originally Posted by deep

Another interesting link https://lkml.org/lkml/2002/6/8/74

That is interesting. It looks like their linux kernel patch adds the "cygwin shell" method of interpreting vfat shortcuts (.lnk files) as symlinks.

The only downside that I can see is that you would need to load your patched kernel with kexec, to avoid interfering with already fragile kindle OTA updates.

It might be safer to add these patches to a shell instead of the kernel, like how cygwin does it. Then do your tcc compiles inside the patched shell.

But really, the USB drive is large enough that we can just use renamed file copies instead of symlinks, which is how I handled this kind of problem in the past. If your library needs three names, just use three copies of that library on your vfat drive. Then you do not need non-standard and potentially fragile vfat symlink support.

[knc1]

Originally Posted by geekmaster

That is interesting. It looks like their linux kernel patch adds the "cygwin shell" method of interpreting vfat shortcuts (.lnk files) as symlinks.

Note the date of that e-mail.
If it hasn't made it into the kernel sources in ten years of waiting, it is unlikely to be added.

[deep]

Originally Posted by geekmaster

That is interesting. It looks like their linux kernel patch adds the "cygwin shell" method of interpreting vfat shortcuts (.lnk files) as symlinks.

The only downside that I can see is that you would need to load your patched kernel with kexec, to avoid interfering with already fragile kindle OTA updates.

It might be safer to add these patches to a shell instead of the kernel, like how cygwin does it. Then do your tcc compiles inside the patched shell.

But really, the USB drive is large enough that we can just use renamed file copies instead of symlinks, which is how I handled this kind of problem in the past. If your library needs three names, just use three copies of that library on your vfat drive. Then you do not need non-standard and potentially fragile vfat symlink support.

Yup ... I guess we could go back to the original problem of interpreting the vfat as a linux partition by KT and a linux partion as a vfat by the PC ...

[deep]

Originally Posted by knc1

Note the date of that e-mail.
If it hasn't made it into the kernel sources in ten years of waiting, it is unlikely to be added.

Yes, It's a really old one ... I thought it might be an interesting way to approach the issue of symlinks... but since it hasn't grabbed the attention of kernel programmers for so many years, it's probably not a good solution ...

[knc1]

Originally Posted by deep

Yes, It's a really old one ... I thought it might be an interesting way to approach the issue of symlinks... but since it hasn't grabbed the attention of kernel programmers for so many years, it's probably not a good solution ...

The Linux Kernel Project has had one (or more) rounds of patent issues over the implementation of VFAT.
Ten years ago feels about right for when they where heavily engaged with MS on the matter.
So this patch might have been ignored due to those troubles, not due to the quality of the patch or the idea.

Note: This is one of the reasons for so many options on the VFAT system.
That allows the selection of a combination of options required to mount most of the other "almost FAT" file systems without trampling on anyone's I.P. rights.

[twobob]

Begin__AntiRant__

For a newb it's always tough to derive the noise from the wheat, the could from the should and the from the

This thread encompasses a range of subjects many of which are intriguing to a novice.

The way they are communicated is far less important than their content.

The threads that contain quality content like this will live on long after flame-fuelled troll drivel has sunk into the necro-abyss of google page 10.

Much appreciate the downloads and the efforts on all sides to drive this tool towards compliance despite all odds , regardless of the individual methodologies, the message is the same:
"Providing compliant tools for the betterment of all"
and that's worth applauding.

So nice one you few brave crew either driven, misguided or just OCD enough to do it.
We salute you.

End__AntiRant__

For my part I'm just relieved to find some vague assistance in getting anything to compile. Happy days. Perhaps I will find a howto step-by-step for slackware, c, arm and native, including all the tool-chains and references.
In the interim this is an excellent tool for trying out some code.

Thanks again.

[knc1]

Originally Posted by twobob

For my part I'm just relieved to find some vague assistance in getting anything to compile. Happy days. Perhaps I will find a howto step-by-step for slackware, c, arm and native, including all the tool-chains and references.
In the interim this is an excellent tool for trying out some code.

Thanks again.

Browse:
https://wiki.mobileread.com/wiki/Tools_Index
Follow:
https://www.mobileread.com/forums/sho...d.php?t=181068
as I finish its write-up. Currently you have to read all the posts.

[geekmaster]

Originally Posted by twobob

So nice one you few brave crew either driven, misguided or just OCD enough to do it.
We salute you.

What worthwhile creation WASN'T from somebody with at least a "little" OCD? Just read a biography of any famous artist, musician, or computer programmer. It takes obsessive compulsion just to make the error-filled compiler output from any significant new program NOT make you run from the room screaming in horror!

[twobob]

Originally Posted by geekmaster

What worthwhile creation WASN'T from somebody with at least a "little" OCD? Just read a biography of any famous artist, musician, or computer programmer. It takes obsessive compulsion just to make the error-filled compiler output from any significant new program NOT make you run from the room screaming in horror!

Seconded. Karma baby

[geekmaster]

@twobob: Welcome to the 2600 club!