Thanks for explaining the incremental behavior of tup.
My question however is on the doubt/trouble I have in understanding these
three phrases in the EXAMPLE of the tup manual:
The foreach :-rule will generate a command to compile each file.
First tup will parse the input section, and use the glob operation on the
database since a '*' is present.
This glob matches foo.c and bar.c.
This is my interpretation of these three phrases:
If the input section of the rule contains a glob on some directory D {
DirNode DN = database.findNode(D)
if (DN == null) DN = database.createDirectoryNode(D)
//Cache the directory content in the tup database
foreach direntry F in D {
FileNode FN = database.findNode(F.name)
if (FN == null) database.createFileNode(F.name)
}
// evaluate the glob on the file nodes in the database
foreach filenode FN in DN: if (glob matches FN.name) inputNodes += FN
} else {
foreach file F in the input section {
FileNode FN = database.findNode(F.name)
if (FN == null) database.createFileNode(F.name)
inputNodes += FN
}
}
// ...continue parsing the output section and cmd sections of the rule
Tup will only re-parse the tupfile when directory D has changed since the
previous build and/or when the tupfile itself has changed.
Another interpretation is:
If the input section of the rule contains a glob on some directory D {
GlobNode DN = database.findNode(D, globPattern)
if (DN == null) DN = database.createGlobNode(D)
//Cache the glob result in the tup database
foreach direntry F in D {
if (glob matches F) {
FileNode FN = database.findNode(F)
if (FN == null) database.createFileNode(F)
inputNodes += FN
}
}
} else {
foreach file F in the input section {
FileNode FN = database.findNode(F)
if (FN == null) database.createFileNode(F)
inputFileNames += FN.name
}
}
This second interpretation however does not reflect the "...and use the
glob operation on the database... "
because it does not apply the glob to the (file nodes in the) database but
directly to the directory entries.
My question: which interpretation best matches tup's implementation? Or is
tup acting in yet another way?
Why do I want to understand this ?
Consider a directory that contains a large numbers files of which only few
files match the glob and
of which only few files will ever be used as input by the build.
For such a directory the first interpretation will result in many not-used
file nodes in the tup database, making the
database larger and possibly slower.
Op zaterdag 30 september 2023 om 05:20:37 UTC+2 schreef [email protected]:
> I'm just another tup user, but I think I can answer your question.
>
> The key thing to understand is that performing a build, with 'tup',
> needn't involve parsing any Tupfiles at all. These files must of course be
> parsed the *first* time. But from that point on, unless they change or the
> listed entries of their local directory change, tup can continue updating
> the outputs of that directory based on updates to the existing inputs using
> nothing but the database, and, as you observe, the cached initial result of
> the globbing.
>
> For larger builds, this distinction of re-parsing vs. re-performing build
> steps becomes obvious from the console output. Re-parsing is the first
> pass, indicated by a yellow progress bar, whereas the second pass--the
> build proper--is indicated with a second green progess bar. Incremental
> builds will for the most part only include the second pass, since fuse can
> reliably inform tup that no re-parse is required.
>
> I hope that's helpful/accurate, it's just what I've gathered from waiting
> on a many incremental builds over the last few years.
>
> On Sat, 30 Sept 2023, 6:07 am Peter Jaspers, <[email protected]> wrote:
>
>> Hi,
>>
>>
>> From the EXAMPLE section in the tup manual:
>> The foreach :-rule will generate a command to compile each file. First
>> tup will parse the input section, and use the glob operation on the
>> database since a '*' is present. This glob matches foo.c and bar.c.
>>
>> To my understanding, for this to work, tup must populate the database
>> with a file node for each file found in the directory *before *it starts
>> parsing the tupfile. Or does tup only create nodes for files that match the
>> glob pattern? In the latter case why then perform a glob operation on the
>> database?
>>
>>
>>
>> Peter
>>
>>
>>
>>
>>
>> --
>> --
>> tup-users mailing list
>> email: [email protected]
>> unsubscribe: [email protected]
>> options: http://groups.google.com/group/tup-users?hl=en
>> ---
>> You received this message because you are subscribed to the Google Groups
>> "tup-users" group.
>> To unsubscribe from this group and stop receiving emails from it, send an
>> email to [email protected].
>> To view this discussion on the web visit
>> https://groups.google.com/d/msgid/tup-users/02686f7a-d99f-4102-aaf9-5dff08ffe750n%40googlegroups.com
>>
>> <https://groups.google.com/d/msgid/tup-users/02686f7a-d99f-4102-aaf9-5dff08ffe750n%40googlegroups.com?utm_medium=email&utm_source=footer>
>> .
>>
>
--
--
tup-users mailing list
email: [email protected]
unsubscribe: [email protected]
options: http://groups.google.com/group/tup-users?hl=en
---
You received this message because you are subscribed to the Google Groups
"tup-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to [email protected].
To view this discussion on the web visit
https://groups.google.com/d/msgid/tup-users/6170c92c-8c96-485c-b951-4603c64cde07n%40googlegroups.com.
